Intelligent production line data fusion method based on cloud edge collaboration

Abstract

The invention provides an intelligent production line data fusion method based on cloud edge collaboration, and relates to the technical field of intelligent manufacturing and industrial internet. In the key production processes of battery cell drying, battery cell assembling and battery pack assembling, an edge end collects key data of surface temperature, voltage and bolt torque of a positive pole piece and a negative pole piece in real time, once abnormity is detected, an alarm is triggered immediately, when the abnormity degree is high, the data is uploaded to a cloud end immediately, and a plan for adjusting key parameters of the production processes is generated in real time. The cloud end gathers the data uploaded by the edge end, carries out statistical analysis on the abnormal frequency of each process in combination with historical data, dynamically adjusts the weight coefficient of each process in quality evaluation according to the abnormal frequency, endows a higher weight coefficient to the process with low abnormal frequency, correspondingly reduces the weight coefficient to the process with high abnormal frequency, and carries out quality evaluation through a quality evaluation formula. And integrating the scores and weights of the processes, and dividing the production quality of the battery into three grades, namely, an excellent grade, a good grade and a poor grade according to a publicity result.

Description

Technical Field

[0001] The present invention relates to the technical fields of intelligent manufacturing and industrial Internet, and particularly to an intelligent production line data fusion method based on cloud-edge collaboration. Background Art

[0002] At present, with the rapid development of intelligent manufacturing and industrial Internet, batteries, as key energy components indispensable in many fields, have attracted increasing attention in terms of their manufacturing processes and quality control. Battery manufacturing covers multiple key processes such as cell drying, cell assembly, and battery Pack assembly. The quality of each process has a decisive impact on the final performance and safety of the battery. In the field of new energy vehicles, battery performance directly determines the vehicle's driving range, power output, and safety performance, and is the core bottleneck for the widespread popularization and technological breakthrough of new energy vehicles.

[0003] In the traditional battery production process, there are many drawbacks in data collection and processing methods. On the one hand, data collection is scattered and lacks systematicness. Data from different production links are often collected by independent devices or systems respectively. These data have inconsistent formats and standards, making data integration extremely difficult. On the other hand, in terms of data processing and analysis, traditional methods mostly rely on manual experience and simple statistical analysis tools. Facing production anomalies such as temperature fluctuations during cell drying, voltage deviations during cell assembly, and unstable bolt torques during battery Pack assembly, it is difficult to monitor and give early warnings in a timely and accurate manner. When anomalies occur, it is often impossible to quickly locate the root cause of the problem and take effective solutions, resulting in low production efficiency, uneven product quality, and even potential safety hazards.

[0004] With the in-depth development of the concept of intelligent manufacturing, cloud-edge collaboration technology has gradually emerged and been applied to the industrial production field. Cloud computing, with its powerful computing power and massive data storage function, can conduct in-depth analysis and complex modeling on large-scale data, providing comprehensive and scientific bases for production decisions. The purpose of the present invention is to propose an intelligent production line data fusion method based on a cloud-edge collaboration architecture, aiming at the key processes of battery manufacturing, to achieve efficient data collection, transmission, fusion, and analysis, thereby effectively improving battery production quality and production efficiency.

[0005] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0006] The purpose of the present invention is to provide an intelligent production line data fusion method based on cloud-edge collaboration to solve the problems raised in the above background art.

[0007] To achieve the above object, the present invention provides the following technical solutions:

[0008] An intelligent production line data fusion method based on cloud-edge collaboration, the specific steps include:

[0009] Step 1: The edge side collects key data in real time, and the key data includes the temperature on the surfaces of the positive and negative electrode plates in the battery cell drying process, the voltage of the battery cell in the battery cell assembly process, and the torque of the bolts in the battery Pack assembly process;

[0010] Step 2: The edge side performs minimum-maximum normalization processing on the key data, constructs a scoring model for each process based on historical data, and judges the degree of abnormality according to the model output result and the threshold of each process scoring model. The degree of abnormality is none, low, and high;

[0011] Step 3: Select no action, the edge side alarms the process, and immediately uploads the data to the cloud while the edge side alarms the process according to the degree of abnormality. After receiving the uploaded data, the cloud automatically generates a plan to adjust the key parameters of the production process;

[0012] Step 4: The data collected by the edge side is regularly stored in the cloud. The cloud constructs a quality evaluation formula according to the scoring model of each process, and realizes dynamic adjustment of the weight coefficient by counting the frequency of abnormal processes to evaluate the product quality.

[0013] Further, in the battery cell drying process, when the positive and negative electrode plates enter the drying link, the edge-side device collects the temperature on the surfaces of the positive and negative electrode plates at a frequency of 1 time per second as set, and records it as T i , where i = 1, 2, 3,... z0, and z0 represents the total number of times of collecting temperature data in the battery cell drying process;

[0014] In the battery cell assembly process, the edge-side device collects the voltage of the battery cell at a frequency of 1 time per 5 seconds as set, and records it as V j , where j = 1, 2, 3,... z1, and z1 represents the total number of times of collecting voltage data in the battery cell assembly process;

[0015] In the battery Pack assembly process, collect the torque data of the battery module fixing bolts, the battery management system connection bolts, the busbar connection bolts, and the cooling system connection bolts respectively. Record the torque data of the battery module fixing bolts as M1 a , where a = 1, 2, 3,... z2, and z2 represents the total number of times of collecting the torque data of the battery module fixing bolts in the battery Pack assembly process;

[0016] Record the torque data of the battery management system connection bolts as M2 b, where b = 1, 2, 3, … z3, and z3 represents the total number of times of collecting the torque data of the battery management system connection bolts in the battery Pack assembly process;

[0017] Record the torque data of the busbar connection bolts as M3 c , where c = 1, 2, 3, … z4, and z4 represents the total number of times of collecting the torque data of the busbar connection bolts in the battery Pack assembly process;

[0018] Record the torque data of the cooling system connection bolts as M4 d , where d = 1, 2, 3, … z5, and z5 represents the total number of times of collecting the torque data of the cooling system connection bolts in the battery Pack assembly process.

[0019] Furthermore, extract the original data X of the key data of each process, the maximum value X of the data max , the minimum value X of the data min , perform min-max normalization processing on the key data:

[0020]

[0021] In the formula, X norm represents the normalized data.

[0022] Furthermore, for the scoring model of the battery cell drying process, first extract the maximum temperature T on the surfaces of the positive and negative electrode plates max , the lowest temperature T min , calculate the temperature fluctuation range:

[0023] ΔT = T max -T min

[0024] In the formula, ΔT represents the temperature fluctuation range on the surfaces of the positive and negative electrode plates in the battery cell drying process;

[0025] Calculate the average temperature:

[0026]

[0027] In the formula, T mean represents the average temperature on the surfaces of the positive and negative electrode plates in the battery cell drying process;

[0028] Calculate the temperature standard deviation:

[0029]

[0030] In the formula, σ T represents the temperature standard deviation on the surfaces of the positive and negative electrode plates in the battery cell drying process;

[0031] Extract data of N' groups of drying processes from historical data, and denote the average temperature of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical temperature fluctuation range:

[0032]

[0033] In the formula, represents the historical temperature fluctuation range;

[0034] The standard deviation of the temperature of the historical drying process is σ Tk , where calculate the average value of the historical temperature standard deviation:

[0035]

[0036] In the formula, represents the average value of the historical temperature standard deviation;

[0037] Construct a scoring model for the battery cell drying process:

[0038]

[0039] In the formula, S T represents the scoring model for the battery cell drying process, α T represents the temperature fluctuation sensitivity coefficient, and α T > 0;

[0040] For the scoring model of the battery cell assembly process, first extract the maximum voltage V of the battery cell max , the minimum voltage V of the battery cell min , and calculate the maximum voltage deviation:

[0041] ΔV = V max -V min

[0042] In the formula, ΔV represents the maximum voltage deviation of the battery cell in the battery cell assembly process;

[0043] Calculate the average voltage:

[0044]

[0045] In the formula, V mean represents the average voltage of the battery cell in the battery cell assembly process;

[0046] Calculate the voltage standard deviation:

[0047]

[0048] In the formula, σ V represents the voltage standard deviation of the battery cell in the battery cell assembly process;

[0049] Extract the data of N' groups of cell assembly processes from the historical data, and denote the average value of the voltage of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical voltage fluctuation amplitude:

[0050]

[0051] In the formula, represents the historical voltage fluctuation amplitude;

[0052] The voltage standard deviation of the historical cell assembly process is σ Vk , and calculate the average value of the historical voltage standard deviation:

[0053]

[0054] In the formula, represents the average value of the historical voltage standard deviation;

[0055] Construct a scoring model for the cell assembly process:

[0056]

[0057] In the formula, S V represents the scoring model of the cell assembly process, α V represents the voltage fluctuation sensitivity coefficient, and α V > 0;

[0058] For the scoring model of the battery Pack assembly process, first calculate the average value of the torque of the fixing bolts of the battery module:

[0059]

[0060] In the formula, M1 mean represents the average value of the torque of the fixing bolts of the battery module;

[0061] Calculate the average value of the torque of the connection bolts of the battery management system:

[0062]

[0063] In the formula, M2 mean represents the average value of the torque of the fixing bolts of the battery module;

[0064] Calculate the average value of the torque of the connection bolts of the busbar:

[0065]

[0066] In the formula, M3 mean represents the average value of the torque of the connection bolts of the busbar;

[0067] Calculate the average value of the torque of the bolts connecting the cooling system:

[0068]

[0069] In the formula, M4 mean represents the average value of the torque of the bolts connecting the cooling system;

[0070] Calculate the standard deviation of the torque of the bolts fixing the battery module:

[0071]

[0072] In the formula, σ M1 represents the standard deviation of the torque of the bolts fixing the battery module;

[0073] Calculate the standard deviation of the torque of the bolts connecting the battery management system:

[0074]

[0075] In the formula, σ M2 represents the standard deviation of the torque of the bolts fixing the battery module;

[0076] Calculate the standard deviation of the torque of the bolts connecting the busbar:

[0077]

[0078] In the formula, σ M3 represents the standard deviation of the torque of the bolts connecting the busbar;

[0079] Calculate the standard deviation of the torque of the bolts connecting the cooling system:

[0080]

[0081] In the formula, σ M4 represents the standard deviation of the torque of the bolts connecting the cooling system;

[0082] Extract the maximum torques M1 max , M2 max , M3 max , M4 max , and the minimum torques M1 min , M2 min , M3 min , M4 min , and calculate the overall torque fluctuation range:

[0083] ΔM = (M1 max - M1 min ) + (M2 max - M2 min ) + (M3 max - M3 min ) + (M4max -M4 min )

[0084] where ΔM represents the overall torque fluctuation amplitude;

[0085] Calculate the average value of the overall torque fluctuation amplitude in the calculation history and the average value of the historical torque standard deviation of the four groups of bolts and Construct a scoring model for the battery Pack assembly process:

[0086]

[0087] where S M represents the scoring model of the battery Pack assembly process.

[0088] Furthermore, set the threshold values of each process scoring model: the threshold values of the cell drying process scoring model are ST1 and ST2 respectively, and ST1 > ST2 > 0; the threshold values of the cell assembly process scoring model are SV1 and SV2 respectively, and SV1 > SV2 > 0; the threshold values of the battery Pack assembly process scoring model are SM1 and SM2 respectively, and SM1 > SM2 > 0;

[0089] When S T > ST1, SV > SV1, S M > SM1, it is determined that the process is normal;

[0090] When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the cell drying process is low; when SV2 < SV ≤ SV1, it is determined that the abnormal degree of the cell assembly process is low; when SM2 < SM ≤ SM1, it is determined that the abnormal degree of the battery Pack assembly process is low;

[0091] When S T ≤ ST2, it is determined that the abnormal degree of the cell drying process is high; when S V ≤ SV2, it is determined that the abnormal degree of the cell assembly process is high; when S M ≤ SM2, it is determined that the abnormal degree of the battery Pack assembly process is high.

[0092] Furthermore, when S T > ST1, sV > SV1, S M > SM1, it is determined that the process is normal and no alarm is required;

[0093] When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the cell drying process is low, and the edge terminal triggers an alarm for the cell drying process;

[0094] When SV2 < SV ≤ SV1, it is determined that the abnormal degree of the battery cell assembly process is low, and the edge terminal triggers an alarm for the battery cell assembly process;

[0095] When SM2 < SM ≤ SM1, it is determined that the abnormal degree of the battery Pack assembly process is low, and the edge terminal triggers an alarm for the battery Pack assembly process;

[0096] When S T ≤ ST2, it is determined that the abnormal degree of the battery cell drying process is high, and the edge terminal triggers an alarm for the battery cell drying process, and the data is immediately uploaded to the cloud;

[0097] When S V ≤ SV2, it is determined that the abnormal degree of the battery cell assembly process is high, and the edge terminal triggers an alarm for the battery cell assembly process, and the data is immediately uploaded to the cloud;

[0098] When S M ≤ SM2, it is determined that the abnormal degree of the battery Pack assembly process is high, and the edge terminal triggers an alarm for the battery Pack assembly process, and the data is immediately uploaded to the cloud.

[0099] Furthermore, when the abnormal degree of the battery cell drying process is high, the cloud extracts the average temperature on the surfaces of the positive and negative electrode plates received, and adjusts the temperature to the historical temperature fluctuation range When the abnormal degree of the battery cell assembly process is high, the cloud extracts the average voltage received until the voltage is adjusted back to the historical voltage fluctuation range When the abnormal degree of the battery Pack assembly process is high, the cloud extracts the torque data of the battery module fixing bolts, the battery management system connection bolts, the busbar connection bolts, and the cooling system connection bolts received, and adjusts the torque values respectively until the average value of the historical overall torque fluctuation range is reached

[0100] Furthermore, extract the scoring models of the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process, and construct a quality evaluation formula:

[0101] Q = ω T ·S T + ω V ·S V + ω M ·S M

[0102] In the formula, Q represents the quality evaluation formula, ω T 、ω V and ω M are weight coefficients.

[0103] Furthermore, set the time window as T s , and count the abnormal frequencies of the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process within the time window:

[0104]

[0105] In the formula, f g represents the abnormal frequency of the g-th process within the time window T s , where g = 1, 2, 3, corresponding to the cell drying process, the cell assembly process, and the battery Pack assembly process respectively, and CHA(g, T s ) represents the number of times with a high degree of recognized abnormality in the g-th process within the time window T s , and N total represents the number of times of the scoring model of the g-th process within the time window T s ;

[0106] The formula for the weight coefficient is:

[0107]

[0108] In the formula, ω h represents each weight coefficient of the quality evaluation formula, ∈ represents a minimum value, and ∈ = 0.001 is taken. Among them, T, V, and M represent the cell drying process, the cell assembly process, and the battery Pack assembly process respectively, and ∑ r∈{t，V，M} 1 / (f r +∈) represents the summation of all processes;

[0109] Dynamically adjust the weight coefficient:

[0110]

[0111] In the formula, ω h (t) represents the weight coefficient of the g-th process in the t-th time window, and ω h (t-1) represents the weight coefficient of the g-th process in the (t - 1)-th time window, that is, the weight coefficient of the previous time window. f g (t) represents the abnormal frequency of the g-th process in the t-th time window, and f r (t) represents the abnormal frequency of the cell drying process, the cell assembly process, and the battery Pack assembly process in the t-th time window. β is a smoothing factor, and its value is 0.85.

[0112] Furthermore, preset the quality evaluation thresholds Q zl1 and Q zl2 , Q zl1 >Q zl2 >0. When Q ≥ Q zl1 , determine that the quality grade is excellent; when Q zl1>Q≥Q zl2 When, the quality grade is determined to be good; when Q zl2 >Q, the quality grade is determined to be poor.

[0113] Compared with the prior art, the beneficial effects of the present invention are:

[0114] The present invention uses the edge side to collect key data such as the surface temperature of the positive and negative electrodes in the battery core drying process, the battery core voltage in the battery core assembly process, and the bolt torque in the battery Pack assembly process in real time. Through min-max normalization processing, the influence of data dimensions is eliminated, and then combined with a large amount of historical data to construct a scoring model for each process, so as to comprehensively monitor and warn whether the production situation is abnormal. When there is a high anomaly, not only does the edge side give an alarm, but also the data is immediately uploaded to the cloud for analysis and use, and it can also guide the adjustment of the production process; the present invention also regularly stores the data in the cloud, and dynamically adjusts the weight coefficient by counting the frequency of abnormal processes, constructs a quality evaluation formula, makes the quality evaluation result more in line with the production reality, and thus optimizes the process. BRIEF DESCRIPTION OF THE DRAWINGS

[0115] Figure 1 It is a schematic diagram of the overall method flow of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0116] In order to make the purpose, technical solution and advantages of the present invention clearer, the following further describes the present invention in detail with reference to specific embodiments.

[0117] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meaning understood by those with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "including" or "comprising" mean that the elements or objects appearing before the word cover the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connected" or "coupled" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Upper", "lower", "left", "right", etc. are only used to represent relative position relationships, and when the absolute position of the object being described changes, the relative position relationship may also change accordingly.

[0118] Embodiment:

[0119] Please refer to Figure 1 , the present invention provides a technical solution:

[0120] An intelligent production line data fusion method based on cloud-edge collaboration, the specific steps include:

[0121] Step 1: The edge device collects key data in real time. The key data includes the temperature on the surfaces of the positive and negative electrodes during the battery cell drying process, the voltage of the battery cell during the battery cell assembly process, and the torque of the bolts during the battery Pack assembly process.

[0122] In the new energy vehicle battery production line, when the positive and negative electrodes enter the drying furnace, the edge device collects the surface temperature at a frequency of once per second through an infrared temperature sensor and records it as T i , where i = 1, 2, 3, … z0, and z0 represents the total number of times of collecting temperature data during the battery cell drying process. Taking the surface temperature of the positive and negative electrodes as important monitored data is because local overheating or insufficient drying will cause problems such as structural defects, increased internal resistance, and material shedding. High-frequency collection can detect these subtle changes in a timely manner;

[0123] In the battery cell assembly process, the edge device collects the voltage of the battery cell at a set frequency of once every 5 seconds and records it as V j , where j = 1, 2, 3, … z1, and z1 represents the total number of times of collecting voltage data during the battery cell assembly process. The battery cell has a relatively stable voltage range at different stages such as charging, discharging, or standing still. If abnormal situations such as sudden increase or decrease in voltage or excessive voltage fluctuation occur, high-frequency collection can capture them in a timely manner, facilitating the operator to quickly judge whether there are problems such as internal short circuit, open circuit, and abnormal electrode materials in the battery cell, avoiding defective battery cells from entering the next process and reducing the generation of defective products;

[0124] In the battery Pack assembly process, torque data of the battery module fixing bolts, battery management system connection bolts, busbar connection bolts, and cooling system connection bolts are collected respectively. Among them, if the torque of the battery module fixing bolts is insufficient, the battery module may loosen due to vibration during vehicle driving or other processes, affecting the stability of the battery system and even possibly resulting in poor connection between modules. If the connection is loose, if the torque of the battery management system connection bolts is insufficient, it may lead to unstable signal transmission and cause the battery management system to malfunction; if the torque of the busbar connection bolts is insufficient, it may generate contact resistance at the busbar connection, resulting in heat generation and even burning of the circuit in severe cases; if the torque of the cooling system connection bolts is insufficient, it may cause leakage of the cooling pipeline, resulting in loss of the cooling medium and reducing the battery performance and life.

[0125] Record the torque data of the battery module fixing bolts as M1 a , where a = 1, 2, 3, … z2, and z2 represents the total number of times of collecting torque data of the battery module fixing bolts during the battery Pack assembly process;

[0126] Record the torque data of the battery management system connection bolts as M2 b, where b = 1, 2, 3, … z3, and z3 represents the total number of times of collecting the torque data of the battery management system connection bolts in the battery Pack assembly process;

[0127] Record the torque data of the busbar connection bolts as M3 c , where c = 1, 2, 3, … z4, and z4 represents the total number of times of collecting the torque data of the busbar connection bolts in the battery Pack assembly process;

[0128] Record the torque data of the cooling system connection bolts as M4 d , where d = 1, 2, 3, … z5, and z5 represents the total number of times of collecting the torque data of the cooling system connection bolts in the battery Pack assembly process.

[0129] Step 2: The edge side performs minimum-maximum normalization processing on the key data, constructs a scoring model for each process based on historical data, and judges the degree of abnormality according to the model output result and the threshold of each process scoring model. The degree of abnormality is none, low, and high;

[0130] In different processes, key data has different dimensions and orders of magnitude. For example, the unit of the surface temperature of the positive and negative electrode plates is degrees Celsius, while the unit of the bolt torque is Newton-meter, and the unit of the cell voltage is volt. Through minimum-maximum normalization processing, these data with different dimensions can be uniformly mapped to a specific interval of [0, 1], making different types of data comparable, facilitating comprehensive analysis and model construction. Extract the original data X of the key data of each process, the maximum value X of the data max , the minimum value X of the data min , perform minimum-maximum normalization processing on the key data:

[0131]

[0132] In the formula, X norm represents the normalized data.

[0133] For the scoring model of the cell drying process, first extract the maximum temperature T on the surface of the positive and negative electrode plates max , the lowest temperature T min , calculate the temperature fluctuation range:

[0134] ΔT = T max -T min

[0135] In the formula, ΔT represents the amplitude of temperature fluctuation on the surfaces of the positive and negative electrode plates during the drying process of the battery cell. The amplitude of temperature fluctuation ΔT can directly reflect the temperature stability of the positive and negative electrode plates during drying. A larger ΔT value indicates larger temperature fluctuations, and there may be local overheating or overcooling, which may affect the drying quality of the positive and negative electrode plates;

[0136] Calculate the average temperature:

[0137]

[0138] In the formula, T mean represents the average temperature on the surfaces of the positive and negative electrode plates during the drying process of the battery cell;

[0139] Calculate the standard deviation of temperature:

[0140]

[0141] In the formula, σ T represents the standard deviation of temperature on the surfaces of the positive and negative electrode plates during the drying process of the battery cell, which can accurately measure the degree of dispersion of the surface temperature of the positive and negative electrode plates relative to the average temperature. A larger σ T value indicates that the temperature distribution is more dispersed, and there may be some areas with too high or too low temperature, which may have an adverse impact on the performance of the positive and negative electrode plates;

[0142] Extract data of N' drying processes from historical data. For example, select data of 100 drying processes. Extracting multiple groups of data can make the statistical results closer to the real situation and reduce sampling errors. Denote the average temperature of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical temperature fluctuation amplitude:

[0143]

[0144] In the formula, represents the historical temperature fluctuation amplitude. By calculating the historical temperature fluctuation amplitude, we can understand the overall temperature fluctuation situation of the battery cell drying process in multiple past batches (groups) from a macroscopic perspective, providing an intuitive indicator for the overall evaluation of the stability of the drying process;

[0145] The standard deviation of temperature of the historical drying process is σ Tk , and the formula for calculating the average value of the historical temperature standard deviation is:

[0146]

[0147] In the formula, represents the average value of the historical temperature standard deviation. By calculating the average value of the historical temperature standard deviation, we can comprehensively reflect the overall degree of dispersion of temperature in multiple drying processes. If A smaller value indicates that during the historical drying process, the temperature was relatively stable with a small fluctuation range; conversely, it indicates that the temperature fluctuated violently and the stability of the drying process was poor.

[0148] Construct a scoring model for the battery cell drying process:

[0149]

[0150] In the formula, S T represents the scoring model for the battery cell drying process, and α T represents the temperature fluctuation sensitivity coefficient, and α T > 0. The temperature fluctuation amplitude reflects the severity of temperature change during the drying process, while the temperature standard deviation reflects the dispersion of temperature data relative to the average value. By comprehensively considering these two indicators, the stability of the temperature during the battery cell drying process can be evaluated comprehensively. At the same time, using the and calculated from historical data as the reference benchmark and comparing the current temperature situation of the drying process with it, the advantages and disadvantages of the current drying process relative to the historical average level can be clearly judged;

[0151] For the scoring model of the battery cell assembly process, first extract the maximum voltage V max of the battery cell and the minimum voltage V min of the battery cell, and calculate the maximum voltage deviation:

[0152] ΔV = V max - V min

[0153] In the formula, ΔV represents the maximum voltage deviation of the battery cell in the battery cell assembly process. A larger maximum voltage deviation indicates that the voltage changes violently during the assembly process, and there may be abnormal situations such as internal short circuits and poor connections; a smaller maximum voltage deviation indicates that the voltage is relatively stable and the battery cell assembly process is relatively reliable;

[0154] Calculate the average voltage:

[0155]

[0156] In the formula, V mean represents the average voltage of the battery cell in the battery cell assembly process and represents the overall voltage level of the battery cell;

[0157] Calculate the voltage standard deviation:

[0158]

[0159] In the formula, σ VIt is expressed as the standard deviation of the voltage of the battery cells in the battery cell assembly process, which measures the degree of dispersion of the voltage data relative to the average value;

[0160] Extract N' groups of data of the battery cell assembly process from the historical data. Denote the average value of the voltage of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical voltage fluctuation amplitude:

[0161]

[0162] In the formula, It is expressed as the historical voltage fluctuation amplitude. By calculating the average value of the voltage fluctuation amplitudes of multiple groups of historical data, it provides a reliable historical reference for the voltage fluctuation assessment of the current battery cell assembly process;

[0163] The standard deviation of the voltage of the historical battery cell assembly process is σ Vk , and calculate the average value of the historical voltage standard deviation:

[0164]

[0165] In the formula, It is expressed as the average value of the historical voltage standard deviation, which synthesizes the voltage standard deviations of multiple groups of historical data and can reflect the degree of voltage dispersion in the battery cell assembly process over a long period of time;

[0166] Construct a scoring model for the battery cell assembly process:

[0167]

[0168] In the formula, S V It is expressed as the scoring model for the battery cell assembly process, α V It is expressed as the voltage fluctuation sensitivity coefficient, and α V >0. The model compares the maximum voltage deviation and the voltage standard deviation with the historical data. The higher the score, the better the voltage stability in the battery cell assembly process and the more reliable the battery cell quality; the lower the score, it indicates that the assembly process needs to be inspected and improved;

[0169] For the scoring model of the battery Pack assembly process, first calculate the average value of the torque of the fixing bolts of the battery module:

[0170]

[0171] In the formula, M1 mean It is expressed as the average value of the torque of the fixing bolts of the battery module;

[0172] Calculate the average value of the torque of the connection bolts of the battery management system:

[0173]

[0174] In the formula, M2 mean represents the average value of the torque of the battery module fixing bolts;

[0175] Calculate the average value of the torque of the busbar connection bolts:

[0176]

[0177] In the formula, M3 mean represents the average value of the torque of the busbar connection bolts;

[0178] Calculate the average value of the torque of the cooling system connection bolts:

[0179]

[0180] In the formula, M4 mean represents the average value of the torque of the cooling system connection bolts. The torques of all bolts used to fix the module should usually be the same. By comparing the average value of the actually measured torque with the standard value, it is possible to intuitively judge whether the quality requirements are met;

[0181] The standard deviation of the torque can quantitatively reflect the degree of dispersion of each group of bolt torque values relative to their average value. The smaller the standard deviation, the closer the torque of each bolt in the group is to the average value, that is, the higher the torque consistency, which means that the bolt fastening effect is more uniform and the force is more balanced.

[0182] Therefore, calculate the standard deviation of the torque of the battery module fixing bolts:

[0183]

[0184] In the formula, σ M1 represents the standard deviation of the torque of the battery module fixing bolts;

[0185] Calculate the standard deviation of the torque of the battery management system connection bolts:

[0186]

[0187] In the formula, σ M2 represents the standard deviation of the torque of the battery module fixing bolts;

[0188] Calculate the standard deviation of the torque of the busbar connection bolts:

[0189]

[0190] In the formula, σ M3 represents the standard deviation of the torque of the busbar connection bolts;

[0191] Calculate the standard deviation of the torque of the cooling system connection bolts:

[0192]

[0193] In the formula, σ M4 represents the standard deviation of the torque of the bolts connecting the cooling system;

[0194] Extract the maximum torques M1 max , M2 max , M3 max , M4 max of the four groups of bolts, and the minimum torques M1 min , M2 min , M3 min , M4 min , and calculate the overall torque fluctuation range:

[0195] ΔM = (M1 max - M1 min ) + (M2 max - M2 min ) + (M3 max - M3 min ) + (M4 max - M4 min )

[0196] In the formula, ΔM represents the overall torque fluctuation range. A smaller overall torque fluctuation range value means that during the assembly process, the tightening degrees of the individual bolts are relatively consistent, indicating a higher stability of the assembly process. Conversely, a larger value indicates that there may be some unstable factors during the assembly process;

[0197] Calculate the average value of the historical overall torque fluctuation range and the average value of the historical torque standard deviation of the four groups of bolts and Construct a scoring model for the battery Pack assembly process:

[0198]

[0199] In the formula, S M represents the scoring model for the battery Pack assembly process. The model comprehensively considers the overall torque fluctuation range and the dispersion degree of the torque of each group of bolts. A higher score means that during the assembly process, the torque fluctuation range of the bolts is smaller, indicating that the assembly process is stable and reliable and can ensure the quality and performance of the battery Pack. Conversely, a lower score indicates that there may be problems with the assembly process and timely adjustment and optimization are required.

[0200] Set the threshold values of the scoring models for each process: Based on the experience in the actual production process and the understanding of each process, evaluate and adjust the threshold values. The threshold values of the scoring model for the cell drying process are ST1 and ST2 respectively, and ST1 > ST2 > 0; the threshold values of the scoring model for the cell assembly process are SV1 and SV2 respectively, and SV1 > SV2 > 0; the threshold values of the scoring model for the battery Pack assembly process are SM1 and SM2 respectively, and SM1 > SM2 > 0;

[0201] When S T > ST1, SV > SV1, S M > SM1, it is determined that the process is normal;

[0202] When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the cell drying process is low; when SV2 < SV ≤ SV1, it is determined that the abnormal degree of the cell assembly process is low; when SM2 < SM ≤ SM1, it is determined that the abnormal degree of the battery Pack assembly process is low;

[0203] When S T ≤ ST2, it is determined that the abnormal degree of the cell drying process is high; when S V ≤ SV2, it is determined that the abnormal degree of the cell assembly process is high; when S M ≤ SM2, it is determined that the abnormal degree of the battery Pack assembly process is high.

[0204] Step 3: Select no action, edge - side trigger process warning for the process, and at the same time, immediately upload the data to the cloud when the edge - side triggers the process warning. When the cloud receives the uploaded data, automatically generate a plan to adjust the key parameters of the production process;

[0205] By setting different scoring thresholds, divide the abnormal situations of each process into three levels: no abnormality, low abnormality, and high abnormality.

[0206] When S T > ST1, SV > SV1, S M > SM1, it is determined that the process is normal and no warning is required;

[0207] When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the cell drying process is low, and the edge - side triggers a warning for the cell drying process;

[0208] When SV2 < S V ≤ SV1, it is determined that the abnormal degree of the cell assembly process is low, and the edge - side triggers a warning for the cell assembly process;

[0209] When SM2 < SM ≤ SM1, it is determined that the abnormal degree of the battery Pack assembly process is low, and the edge - side triggers a warning for the battery Pack assembly process;

[0210] For example, in the case of a low anomaly, the edge device promptly triggers an alarm for the corresponding process, enabling on-site operators to quickly learn about potential minor issues in the process. This allows them to conduct preliminary inspections and make adjustments in a timely manner, preventing the problems from worsening. In the battery cell drying process, when the result of the scoring model indicates a low level of anomaly, the operator can promptly check whether there are small fluctuations in the operating parameters of the drying equipment and take timely measures to correct them, preventing impacts on subsequent production processes.

[0211] When S T ≤ ST2, it is determined that the anomaly level of the battery cell drying process is high. The edge device triggers an alarm for the battery cell drying process, and the data is immediately uploaded to the cloud.

[0212] When S V ≤ SV2, it is determined that the anomaly level of the battery cell assembly process is high. The edge device triggers an alarm for the battery cell assembly process, and the data is immediately uploaded to the cloud.

[0213] When S M ≤ SM2, it is determined that the anomaly level of the battery Pack assembly process is high. The edge device triggers an alarm for the battery Pack assembly process, and the data is immediately uploaded to the cloud.

[0214] When the model scoring result indicates a high level of anomaly, not only does the edge device trigger an alarm, but the data is also immediately uploaded to the cloud. This ensures that once a serious anomaly occurs, both on-site and remote managers can quickly obtain information. The cloud can aggregate various professional knowledge and data resources, enabling a more comprehensive analysis and diagnosis of high anomalies more quickly and guiding on-site personnel to take effective solutions. For example, when a high anomaly occurs in the battery Pack assembly process, the cloud can call historical data and experts to quickly determine the root cause of the problem and provide accurate solutions for the on-site personnel, minimizing production losses to the greatest extent.

[0215] When the anomaly level of the battery cell drying process is high, after the cloud extracts the average surface temperature of the positive and negative electrode plates received, it can analyze the difference between the current average surface temperature of the positive and negative electrode plates and the historical temperature fluctuation range, determine whether the temperature is too high or too low, and the approximate direction and amplitude of adjustment. For example, according to the type and characteristics of the drying equipment, select an appropriate way to increase or decrease the temperature. If it is an electric heating drying equipment, the heating power can be adjusted to increase or decrease the temperature; if it is a drying equipment using hot air circulation, the flow rate and temperature of the hot air can be adjusted to control the temperature inside the drying equipment.

[0216] When the abnormality degree of the battery cell assembly process is high, the cloud extracts the average voltage received, compares the current average voltage with the historical voltage fluctuation range, determines whether the voltage is too high or too low, and the direction and approximate range for adjusting the current. For example, the current can be adjusted by adjusting the output voltage of the power supply, changing the resistance in the circuit, or using a current adjustment device until the voltage of the battery cell returns to the historical voltage fluctuation range.

[0217] When the abnormality degree of the battery Pack assembly process is high, the cloud extracts the torque data of the fixing bolts of the battery module, the connecting bolts of the battery management system, the connecting bolts of the busbar, and the connecting bolts of the cooling system, calculates the average torque of each bolt group and the overall torque fluctuation range respectively, and compares with the average value of the historical overall torque fluctuation range to determine which bolt groups need to adjust the torque and the direction and amplitude of the adjustment until the average value of the historical overall torque fluctuation range is reached.

[0218] Step 4: The data collected by the edge side is regularly stored in the cloud. The cloud constructs a quality evaluation formula based on the scoring model of each process, and realizes the dynamic adjustment of the weight coefficient by counting the frequency of abnormal processes to evaluate the product quality.

[0219] Extract the scoring models of the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process respectively, and construct a quality evaluation formula:

[0220] Q = ω T ·S T + ω V ·S V + ω M ·S M

[0221] In the formula, Q represents the quality evaluation formula, ω T 、ω V and ω M are weight coefficients, which are used to reflect the relative importance of each process in the overall quality evaluation. The score of each process can intuitively reflect the quality status of the process at a specific time. Different processes have different impacts on the final quality of the battery. Through the quality evaluation formula, the quality of each process can be quantified.

[0222] Set the time window as T s , and count the abnormal frequencies of the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process within the time window:

[0223]

[0224] In the formula, f g represents the time window T sThe abnormal frequency of the g-th process, where g = 1, 2, 3, corresponding to the cell drying process, the cell assembly process, and the battery Pack assembly process respectively, CHA(g, T s ) represents the number of times with a high degree of recognized abnormality in the g-th process within the time window T s , N total represents the number of times of the scoring model in the g-th process within the time window T s . The abnormal frequency can reflect the probability of serious abnormalities occurring in each process over a period of time and is an important indicator for measuring process stability;

[0225] The formula for the weight coefficient is:

[0226]

[0227] In the formula, ω h represents each weight coefficient of the quality assessment formula, ∈ represents a minimum value, taking ∈ = 0.001 to avoid the case of a zero denominator. Among them, T, V, M represent the cell drying process, the cell assembly process, and the battery Pack assembly process respectively, and Σ r∈{T，V，M} 1 / (f r +∈) represents the summation of all processes , which is used here to calculate the initial weight coefficient of each process.

[0228] The lower the abnormal frequency, the more stable the process, and the relatively more reliable its impact on the overall quality, and a higher weight can be assigned; conversely, for a process with a high abnormal frequency, its stability is poor, and the weight should be relatively low. In this way, the importance of each process in the comprehensive evaluation can be dynamically adjusted according to the actual operation situation of each process.

[0229] Dynamic adjustment of the weight coefficient:

[0230]

[0231] In the formula, ω h (t) represents the weight coefficient of the g-th process in the t-th time window, ω h (t-1) represents the weight coefficient of the g-th process in the (t - 1)-th time window, that is, the weight coefficient of the previous time window, fg (t) represents the abnormal frequency of the g-th process in the t-th time window, f r (t) represents the abnormal frequency of the cell drying process, the cell assembly process, and the battery Pack assembly process in the t-th time window, and β is a smoothing factor with a value of 0.85.

[0232] The production process is dynamically changing, and the stability of each process may change over time. By dynamically adjusting the weight coefficients, such changes can be reflected in a timely manner, making the quality assessment results more adaptable to the actual production situation. The reason for the smoothing factor β = 0.85 is that when calculating the weight coefficients of the current time window, higher values will be assigned to the weight coefficients of the historical time windows. This is because the historical weight coefficients reflect the relative importance of each process over a past period of time, avoiding large fluctuations in the weight coefficients due to abnormal frequencies within individual time windows.

[0233] If the abnormal frequency of a certain process increases and the output result of its scoring model is low, an increase in weight will make its negative impact on the quality assessment formula greater, resulting in a decrease in the result. Therefore, a quality assessment threshold Q zl1 and Q zl2 are preset. The specific values are set in combination with the enterprise's product quality objectives. Among them, Q zl1 >Q zl2 >0. When Q ≥ Q zl1 , the quality grade is determined to be excellent; when Q zl1 >Q ≥ Q zl2 , the quality grade is determined to be good; when Q zl2 >Q, the quality grade is determined to be poor.

[0234] The above formulas are all dimensionless and take their numerical values for calculation. The formula is obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formula are set by technicians in this field according to the actual situation.

[0235] The above embodiments can be implemented in whole or in part through software, hardware, firmware, or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented through electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0236] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0237] The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application.

Claims

1. An intelligent production line data fusion method based on cloud-edge collaboration, characterized in that The specific steps include: Step 1: The edge device collects key data in real time. The key data includes the temperature on the surfaces of the positive and negative electrodes during the battery cell drying process, the voltage of the battery cell during the battery cell assembly process, and the torque of the bolts during the battery Pack assembly process. Step 2: The edge device performs min-max normalization on the key data, constructs a scoring model for each process based on historical data, and determines the degree of abnormality according to the model output result and the threshold of each process scoring model. The degrees of abnormality are none, low, and high. Step 3: According to the degree of abnormality, select no action, the edge device alarms for the process, and while the edge device alarms for the process, the data is immediately uploaded to the cloud. When the cloud receives the uploaded data, it automatically generates a plan to adjust the key parameters of the production process. Step 4: The data collected by the edge device is regularly stored in the cloud. The cloud constructs a quality evaluation formula based on the scoring model of each process, and realizes the dynamic adjustment of the weight coefficient by counting the frequency of abnormal processes to evaluate the product quality.

2. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 1, wherein: In Step 1, the edge device's real-time collection of key data includes the following steps: In the battery cell drying process, when the positive and negative electrode sheets enter the drying stage, the edge device collects the temperature on the surface of the positive and negative electrode sheets at a set frequency of once per second and records it as T i , where i = 1, 2, 3, … z0, and z0 represents the total number of times of collecting temperature data in the battery cell drying process; During the battery cell assembly process, the edge device collects the voltage of the battery cell at a set frequency of once every 5 seconds and records it as V j , j = 1, 2, 3, … z1, where z1 represents the total number of times the voltage data is collected during the battery cell assembly process; In the battery Pack assembly process, torque data of the fixing bolts of the battery modules, the connecting bolts of the battery management system, the connecting bolts of the busbars, and the connecting bolts of the cooling system are collected respectively, and the torque data of the fixing bolts of the battery modules are recorded as M1 a , where a = 1, 2, 3,..., z2, and z2 represents the total number of times of collecting the torque data of the fixing bolts of the battery modules in the battery Pack assembly process; Record the torque data of the battery management system connection bolts as M2 b , where b = 1, 2, 3, … z3, and z3 represents the total number of times of collecting the torque data of the battery management system connection bolts in the battery Pack assembly process; Record the torque data of the busbar connection bolts as M3 c , where c = 1, 2, 3, … z4, and z4 represents the total number of times the torque data of the busbar connection bolts is collected in the battery Pack assembly process; Record the torque data of the cooling system connection bolts as M4 d , where d = 1, 2, 3, … z5, and z5 represents the total number of times of collecting the torque data of the cooling system connection bolts in the battery Pack assembly process.

3. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 2, wherein: In Step 2, the method for the edge device to perform min-max normalization on the key data is: Extract the original data X of the key data for each process, the maximum value X of the data max , the minimum value X of the data min , perform minimum-maximum normalization processing on the key data: where X norm represents the normalized data.

4. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 2, wherein: In Step 2, the method for constructing a scoring model for each process based on historical data is: For the scoring model of the battery cell drying process, first extract the maximum temperature T on the surfaces of the positive and negative electrode sheets max , the minimum temperature T min , and calculate the temperature fluctuation range: ΔT = T max - T min In the formula, ΔT represents the fluctuation amplitude of the temperature on the surfaces of the positive and negative electrodes during the battery cell drying process; Calculate the average temperature: Wherein, T mean represents the average surface temperature of the positive and negative electrode plates during the drying process of the battery cell; Calculate the standard deviation of the temperature: Where, σ T represents the standard deviation of the surface temperature of the positive and negative electrode plates during the drying process of the battery cell; Extract data of N' groups of drying processes from historical data, and denote the average value of the temperature of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical temperature fluctuation range: In the formula, represents the historical temperature fluctuation range; The standard deviation of the temperature of the historical drying process is σ Tk , where the average value of the historical temperature standard deviation is calculated: In the formula, represents the average value of the historical temperature standard deviation; Construct a scoring model for the battery cell drying process: Where S T represents the scoring model for the cell drying process, and α T represents the temperature fluctuation sensitivity coefficient, and α T > 0; For the scoring model of the battery cell assembly process, first extract the maximum voltage V of the battery cell max , the minimum voltage V min , and calculate the maximum voltage deviation: ΔV = V max -V min In the formula, ΔV represents the maximum deviation of the voltage of the battery cell during the battery cell assembly process; Calculate the average voltage: Where, V mean represents the average voltage of the battery cell in the battery cell assembly process; Calculate the standard deviation of the voltage: where α V represents the standard deviation of the voltage of the battery cell in the battery cell assembly process; Extract data of N' groups of battery cell assembly processes from historical data, and denote the average value of the voltage of the k-th group as where k = 1, 2, 3, …, N', and calculate the historical voltage fluctuation amplitude: wherein, is expressed as the historical voltage fluctuation amplitude; The voltage standard deviation of the historical battery cell assembly process is σ Vk , calculate the average value of the historical voltage standard deviation: Wherein, is expressed as the average value of the historical voltage standard deviation; Construct a scoring model for the battery cell assembly process: Where S V represents the scoring model for the cell assembly process, and α V represents the voltage fluctuation sensitivity coefficient, and α V > 0; For the scoring model of the battery Pack assembly process, first calculate the average torque of the bolts for fixing the battery modules: Wherein, M1 mean represents the average value of the torque of the battery module fixing bolts; Calculate the average torque of the bolts connecting the battery management system: Wherein, M2 mean represents the average value of the torque of the battery module fixing bolts; Calculate the average torque of the bolts connecting the busbars: where M3 mean represents the average value of the torque of the busbar connection bolts; where M4 mean represents the average value of the torque of the bolts connecting the cooling system; Calculate the average torque of the bolts connecting the cooling system: where σ M1 represents the standard deviation of the torque of the battery module fixing bolts; Where, σ M2 represents the standard deviation of the torque of the battery module fixing bolts; Calculate the standard deviation of the torque of the bolts for fixing the battery modules: where, σ M3 represents the standard deviation of the torque of the busbar connection bolts; where, σ M4 represents the standard deviation of the torque of the bolts connecting the cooling system; Extract the maximum torque M1 of the four groups of bolts max , M2 max , M3 max , M4 max , the minimum torque M1 min , M2 min , M3 min , M4 min , calculate the overall torque fluctuation range: ΔM=(M1 max -M1 min )+(M2 max -M2 min )+(M3 max -M3 min )+(M4 max -M4 min ) Calculate the standard deviation of the torque of the bolts connecting the battery management system: Calculate the average value of the overall torque fluctuation amplitude of the calculation history and the average value of the historical torque standard deviation of the four groups of bolts and Construct a scoring model for the battery Pack assembly process: Where S M represents the scoring model for the battery Pack assembly process.

5. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 4, wherein: Calculate the standard deviation of the torque of the bolts connecting the busbars: When S T > ST1, SV > SV1, S M > SM1, it is determined that the process is normal; When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the battery cell drying process is low; when SV2 < SV ≤ SV1, it is determined that the abnormal degree of the battery cell assembly process is low; when SM2 < SM ≤ SM1, it is determined that the abnormal degree of the battery Pack assembly process is low; When S T ≤ ST2, it is determined that the abnormal degree of the battery cell drying process is high; when S V ≤ SV2, it is determined that the abnormal degree of the battery cell assembly process is high; when S M ≤ SM2, it is determined that the abnormal degree of the battery Pack assembly process is high.

6. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 5, wherein: When S T > ST1, SV > SV1, S M > SM1, it is determined that the process is normal and no alarm is required; When ST2 < S T ≤ ST1, it is determined that the abnormal degree of the battery cell drying process is low, and the edge end triggers an alarm for the battery cell drying process; Calculate the standard deviation of the torque of the bolts connecting the cooling system: When S T ≤ ST2, it is determined that the abnormal degree of the battery cell drying process is high, and the edge end triggers an alarm for the battery cell drying process, and the data is immediately uploaded to the cloud; When S V ≤ SV2, it is determined that the abnormal degree of the battery cell assembly process is high, and the edge end triggers an alarm for the battery cell assembly process, and the data is immediately uploaded to the cloud; When S M ≤ SM2, it is determined that the abnormal degree of the battery Pack assembly process is high, and the edge side triggers an alarm for the battery Pack assembly process, and the data is immediately uploaded to the cloud.

7. A data fusion method for an intelligent production line based on cloud-edge collaboration according to claim 4, characterized in that: In the formula, ΔM represents the overall torque fluctuation amplitude; In Step 2, the method for determining the degree of abnormality according to the model output result and the threshold of each process scoring model is: Set the thresholds of each process scoring model: The thresholds of the scoring model for the battery cell drying process are ST1 and ST2 respectively, and ST1>ST2>0; the thresholds of the scoring model for the battery cell assembly process are SV1 and SV2 respectively, and SV1>SV2>0; the thresholds of the scoring model for the battery Pack assembly process are SM1 and SM2 respectively, and SM1>SM2>0; In Step 3, selecting no action, the edge device alarming for the process, and while the edge device alarms for the process, the data is immediately uploaded to the cloud includes the following steps: When SV2 < SV ≤ SV1, it is determined that the degree of abnormality of the battery cell assembly process is low, and the edge device triggers an alarm for the battery cell assembly process; When SM2 < SM ≤ SM1, it is determined that the degree of abnormality of the battery Pack assembly process is low, and the edge device triggers an alarm for the battery Pack assembly process; In Step 3, generating a plan to adjust the key parameters in the production process includes the following steps: When the abnormality degree of the battery cell drying process is high, the cloud extracts the average temperature of the positive and negative electrode surfaces received and adjusts the temperature to the historical temperature fluctuation range When the abnormality degree of the battery cell assembly process is high, the cloud extracts the average voltage received until the voltage is adjusted back to the historical voltage fluctuation range When the abnormality degree of the battery Pack assembly process is high, the cloud extracts the torque data of the battery module fixing bolts, battery management system connection bolts, busbar connection bolts, and cooling system connection bolts received, and adjusts the torque values respectively until the average value of the historical overall torque fluctuation range is reached 8. A data fusion method for an intelligent production line based on cloud-edge collaboration according to claim 4, characterized in that: In step 4, the method for the cloud to construct a quality evaluation formula based on the scoring models of each process is as follows: Extract the scoring models of the cell drying process, cell assembly process, and battery Pack assembly process, and construct a quality evaluation formula: Q = ω T ·S T + ω V ·S V + ω M ·S M In the formula, Q represents the quality assessment formula, ω T , ω V and ω M are weighting coefficients.

9. The intelligent production line data fusion method based on cloud-edge collaboration according to claim 8, wherein: In step 4, the method for dynamically adjusting the weight coefficient by statistically counting the abnormal process frequency is as follows: Set the time window as T s , and count the abnormal frequencies of the cell drying process, the cell assembly process, and the battery Pack assembly process within the time window: where f g represents the abnormal frequency of the g-th process within the time window T s , where g = 1, 2, 3, corresponding to the cell drying process, the cell assembly process, and the battery Pack assembly process respectively, and CHA(g, T s ) represents the number of times with a high degree of recognized abnormality in the g-th process within the time window T s , and N total represents the number of times of the scoring model for the g-th process within the time window T s ; The formula for the weight coefficient is: where ω h represents each weight coefficient of the quality evaluation formula, ∈ represents a minimum value, and ∈ = 0.001 is taken. Among them, T, V, and M respectively represent the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process, and Σ r∈{T，V，M} 1 / (f r +∈) represents the summation of all processes; Dynamically adjust the weight coefficient: where ω h (t) represents the weight coefficient of the g-th process in the t-th time window, and ω h (t-1) represents the weight coefficient of the g-th process in the (t - 1)-th time window, that is, the weight coefficient of the previous time window, and f g (t) represents the abnormal frequency of the g-th process in the t-th time window, and f r (t) represents the abnormal frequencies of the battery cell drying process, the battery cell assembly process, and the battery Pack assembly process in the t-th time window. β is a smoothing factor with a value of 0.

85.

10. A data fusion method for an intelligent production line based on cloud-edge collaboration according to claim 9, characterized in that: In step 4, the method for evaluating the quality of the product is as follows: Preset the quality assessment threshold Q zl1 and Q zl2 , Q zl1 >Q zl2 >0, when Q ≥ Q zl1 , determine that the quality level is excellent; when Q zl1 >Q ≥ Q zl2 , determine that the quality level is good; when Q zl2 >Q, determine that the quality level is poor.

Images