Improved manufacturing method for head of core rod of PQF continuous rolling unit

Abstract

The invention relates to the technical field of core rod machining, and discloses a PQF continuous rolling unit core rod head improved manufacturing method which comprises the following steps of S1, blank screening, selecting a core rod blank matched with a PQF continuous rolling unit, S2, rough machining, S3, head finish turning, S4, chromium plating treatment, S5, dehydrogenation treatment and S6, finished product inspection. By redesigning the head of the core rod, the problem of scratches is solved, production efficiency is improved, as the scratches on the head of the core rod disappear, pipe inferior-quality products caused by scratching the inner surface of a pipe are reduced, reworking or scrapping treatment of the inferior-quality products is avoided, material waste and time loss in the production process are reduced, and the production cost is reduced. Due to the stable core rod head quality, the number of times of shutdown inspection and equipment adjustment caused by the quality problem is reduced, a continuous rolling unit can continuously and stably operate, the pipe yield in unit time is improved, and the comprehensive production cost is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of mandrel processing, and particularly to an improved manufacturing method for the head of a mandrel of a PQF continuous rolling mill. Background Art

[0002] The PQF continuous rolling mill is an advanced equipment widely used in the field of pipe production. Through the multi-stand continuous rolling process, it realizes the efficient and high-precision rolling of pipes. During the PQF continuous rolling process, the mandrel plays a crucial role. It is located inside the pipe, provides support and guidance for the rolling of the pipe, and directly affects the inner surface quality and dimensional accuracy of the pipe.

[0003] In the prior art, the conical horizontal length of the head of the mandrel used in the PQF continuous rolling mill is 175.1 mm, and the angle is 30°. During production, the mandrel is placed inside the pipe and enters the continuous rolling mill synchronously with the pipe. During the continuous rolling process, multiple stands sequentially apply rolling forces to the pipe, causing the pipe to undergo plastic deformation under the support of the mandrel.

[0004] Although the existing mandrel head design meets the basic production requirements of the PQF continuous rolling mill to a certain extent, obvious defects are still exposed in practical applications. Due to the design limitations of the conical horizontal length and angle of the head, scratches are extremely likely to occur on the mandrel head during the rolling process. These scratches will leave marks on the inner surface of the pipe, seriously affecting the inner surface quality of the pipe, resulting in an uneven surface of the pipe, reducing the service performance and reliability of the pipe. Moreover, scratched pipes often require additional inspection, repair or scrapping, increasing production costs and reducing production efficiency. In view of this, we propose an improved manufacturing method for the head of a mandrel of a PQF continuous rolling mill. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides an improved manufacturing method for the head of a mandrel of a PQF continuous rolling mill, which solves the problem that scratches are extremely likely to occur on the mandrel head during the rolling process due to the design limitations of the conical horizontal length and angle of the head.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: An improved manufacturing method for the head of a mandrel of a PQF continuous rolling mill includes the following steps:

[0007] S1: Blank Screening Step

[0008] Select a mandrel blank suitable for the PQF continuous rolling mill, and comprehensively detect the material, size and physical properties of the blank;

[0009] S2: Rough Machining Step

[0010] Perform preliminary turning on the blank to remove surface impurities and surplus, and prepare for subsequent finish turning;

[0011] S3: Head finish turning step

[0012] Turn the head of the mandrel to a conical angle with a horizontal length of 300 mm, and turn a cone with an angle of 12° between the distances of 100 mm - 300 mm;

[0013] S4: Chrome plating treatment step

[0014] Chrome plate the turned mandrel to make the chrome plating layer have a specific thickness and hardness;

[0015] S5: Dehydrogenation treatment step

[0016] Perform dehydrogenation operation on the chrome-plated mandrel to eliminate internal stress;

[0017] S6: Finished product inspection step

[0018] Inspect multiple indicators including dimensions, hardness, and surface quality of the head of the treated mandrel.

[0019] Preferably, in the S1 blank screening step, the material of the mandrel blank is chrome molybdenum alloy steel, in which the chromium content is 2.0% - 2.5%, the molybdenum content is 0.8% - 1.2%, the carbon content is 0.3% - 0.5%, and the rest is iron and inevitable impurities; the blank size tolerance is controlled within ±0.5 mm, and the hardness range is 200 - 220 HBW.

[0020] Preferably, in the S2 rough machining step, a lathe is used for turning, the spindle speed of the lathe is 200 - 300 r / min, the feed rate is 0.2 - 0.3 mm / r, and the cutting depth is 1 - 2 mm; the turning tool is selected as a YT15 carbide tool, the rake angle of the tool is 10° - 15°, and the clearance angle is 6° - 8°.

[0021] Preferably, a surface defect detection system based on machine vision and deep learning algorithm is introduced between the S2 rough machining step and the S3 head finish turning step. The system uses an industrial camera to collect the surface image of the mandrel, and through a convolutional neural network, scratches, cracks, and sand hole defects in the image are recognized and classified in real time. A defect size threshold is set. Once a defect exceeding the threshold is detected, the subsequent processing parameters are automatically adjusted or the mandrel is marked.

[0022] Preferably, in the S3 head finish turning step, a high-precision CNC lathe is used, the spindle speed is 300 - 400 r / min, the feed rate is 0.05 - 0.1 mm / r, and the cutting depth is 0.1 - 0.2 mm; during the turning process, coolant is used for cooling, and the coolant is emulsion with a concentration of 5% - 8%. A surface defect detection system based on machine vision and deep learning algorithm is introduced between the rough machining step and the head finish turning step.

[0023] Preferably, in the S3 head finish turning step, an adaptive turning parameter optimization algorithm is used. This algorithm is based on the deep deterministic policy gradient algorithm of reinforcement learning, which collects environmental state data including tool wear degree, cutting force magnitude, and workpiece surface roughness in real time, and dynamically adjusts the spindle speed, feed rate, and cutting depth of the lathe.

[0024] Preferably, in the S4 chrome plating treatment step, the chrome plating electrolyte formula is: chromic anhydride 250 - 300 g / L, sulfuric acid 2 - 2.5 g / L, additive 0.5 - 1 g / L; the chrome plating temperature is 50 - 55 °C, the electroplating current density is 30 - 35 A / dm2, and the chrome plating time is 2 - 3 hours to form a chrome plating layer with a thickness of 0.045 - 0.055 mm and a hardness of 60 - 62 HRC.

[0025] Preferably, in the S5 dehydrogenation treatment step, the mandrel is placed in a heating furnace and heated to 200 - 250 °C at a heating rate of 50 - 80 °C / h, held for 3 - 5 hours, and then cooled to room temperature at a cooling rate of 30 - 50 °C / h.

[0026] Preferably, in the S6 finished product inspection step, a coordinate measuring instrument is used to detect that the tolerance of the horizontal length of the conical angle at the head of the mandrel is within ±0.1 mm, and the tolerance of the conical angle between 100 mm - 300 mm is within ±0.1 °; a Rockwell hardness tester is used to detect the hardness of the chrome plating layer, and the hardness deviation is within ±1 HRC; a surface roughness instrument is used to detect that the surface roughness Ra value is not greater than 0.8 μm.

[0027] Preferably, after the S6 finished product inspection step, a quality traceability and continuous improvement system based on the association rule mining algorithm is adopted. This algorithm conducts association rule mining on all data in the manufacturing process, and through the algorithm, finds the key factor combinations affecting the quality of the mandrel. According to the mined association rules, a quality traceability system is established.

[0028] The present invention provides an improved manufacturing method for the head of a PQF continuous rolling mill mandrel. It has the following beneficial effects:

[0029] 1. By redesigning the head of the mandrel, the present invention not only eliminates the problem of scoring, but also improves the production efficiency. Since the scoring on the head of the mandrel disappears, the number of defective pipes caused by scratching the inner surface of the pipe is reduced, avoiding the rework or scrapping of defective pipes, reducing material waste and time loss in the production process. Moreover, the stable quality of the mandrel head reduces the number of downtime inspections and equipment adjustments caused by quality problems, enabling the continuous and stable operation of the continuous rolling mill, increasing the pipe production per unit time, and effectively reducing the comprehensive production cost.

[0030] 2. The present invention brings many significant benefits to the core rod head manufacturing of the PQF continuous rolling mill through an adaptive turning parameter optimization algorithm. In terms of improving product quality, it can sense tool wear, cutting force and changes in workpiece surface roughness in real time, accurately adjust lathe parameters, effectively reduce scratches on the core rod head, avoid scratching the inner surface of the pipe, and greatly improve the inner surface quality of the pipe. From the perspective of production efficiency, the algorithm optimizes parameters according to working conditions, reduces processing time, and at the same time reduces tool loss, reduces the frequency of tool changes, and reduces downtime, thereby significantly improving overall production efficiency.

[0031] 3. The quality traceability system established by the present invention can mine the association rules between manufacturing data and accurately locate the root causes of quality problems, including abnormalities in blank composition and processing parameters. Based on this, enterprises can optimize manufacturing processes, accumulate knowledge and experience, improve production management levels, warn of risks in advance, and reasonably allocate resources, thereby improving product quality, reducing defective rates, and enhancing market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of the improved manufacturing method of the mandrel head of the PQF continuous rolling mill;

[0033] Figure 2 It is a schematic diagram of the adaptive turning parameter optimization algorithm of the present invention;

[0034] Figure 3 Schematic diagram of the quality tracing and continuous improvement system based on the association rule mining algorithm of the present invention. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Example:

[0037] Please see attached Figure 1 - Attachment Figure 3 The embodiment of the present invention provides an improved manufacturing method for a mandrel head of a PQF continuous rolling mill, comprising the following steps:

[0038] S1: Billet screening step

[0039] Select the mandrel billet suitable for the PQF continuous rolling mill and conduct a comprehensive inspection of the billet material, size and physical properties;

[0040] S2: Roughing step

[0041] Perform preliminary turning on the blank to remove surface impurities and surplus, preparing for subsequent finish turning.

[0042] S3: Head finish turning step

[0043] Turn the head of the mandrel to a conical angle horizontal length of 300 mm, and turn a cone with an angle of 12° between the distances of 100 mm - 300 mm.

[0044] S4: Chrome plating treatment step

[0045] Chrome plate the turned mandrel to make the chrome plating layer have a specific thickness and hardness.

[0046] S5: Dehydrogenation treatment step

[0047] Perform dehydrogenation operation on the chrome plated mandrel to eliminate internal stress.

[0048] S6: Finished product inspection step

[0049] Inspect the head of the treated mandrel for multiple indicators including dimensions, hardness, and surface quality.

[0050] In the S1 blank screening step, the material of the mandrel blank is chrome molybdenum alloy steel, where the chromium content is 2.0% - 2.5%, the molybdenum content is 0.8% - 1.2%, the carbon content is 0.3% - 0.5%, and the rest is iron and inevitable impurities; the blank size tolerance is controlled within ±0.5 mm, and the hardness range is 200 - 220 HBW.

[0051] In the S2 rough machining step, use a lathe for turning. The lathe spindle speed is 200 - 300 r / min, the feed rate is 0.2 - 0.3 mm / r, and the cutting depth is 1 - 2 mm; the turning tool is selected as a YT15 carbide tool, with the tool rake angle being 10° - 15° and the clearance angle being 6° - 8°.

[0052] Introduce a surface defect detection system based on machine vision and deep learning algorithms between the S2 rough machining step and the S3 head finish turning step. The system uses an industrial camera to collect the surface image of the mandrel, and uses a convolutional neural network to perform real-time identification and classification of scratches, cracks, and sand holes in the image. Set the defect size threshold. Once a defect exceeding the threshold is detected, automatically adjust the subsequent processing parameters or mark the mandrel.

[0053] In the S3 head finish turning step, a high-precision CNC lathe is used, with the spindle speed being 300 - 400 r / min, the feed rate being 0.05 - 0.1 mm / r, and the cutting depth being 0.1 - 0.2 mm; during the turning process, coolant is used for cooling, the coolant is emulsion with a concentration of 5% - 8%, and a surface defect detection system based on machine vision and deep learning algorithm is introduced between the rough machining and the head finish turning step.

[0054] In the S3 head finish turning step, an adaptive turning parameter optimization algorithm is applied. This algorithm is based on the deep deterministic policy gradient algorithm of reinforcement learning, which collects environmental state data including tool wear degree, cutting force magnitude, and workpiece surface roughness in real time, and dynamically adjusts the spindle speed, feed rate, and cutting depth of the lathe. Here, the following algorithm is established:

[0055] Step 1: Environment definition and state, action, reward design

[0056] State definition: Normalize the data such as tool wear degree, cutting force magnitude, and workpiece surface roughness as state s;

[0057] Action definition: Action a is the spindle speed, feed rate, and cutting depth of the lathe, each with a value range;

[0058] Reward design: Reward function r = α·R quality +β·R efficiency , where α and β are weight coefficients, and R q uality is the turning quality reward, and R efficiency is the machining efficiency reward;

[0059] Step 2: Construct the actor network and the critic network

[0060] Actor network: Input state s, output action a, using the MLP structure with parameters θ μ ;

[0061] Critic network: Input state s and action a, output Q value, using the MLP structure with parameters θ Q ;

[0062] Step 3: Initialize the experience replay buffer and the target network

[0063] The experience replay buffer is used to store experience data (s t ,a t ,r t ,s t+1 );

[0064] Target actor network μ′(s|θ μ′ ) and target critic network Q′(s,a|θ Q′ ), initially θμ′ = θ μ , θ Q′ = θ Q ;

[0065] Step 4: The agent interacts with the environment and stores experiences

[0066] Initialize the state s0;

[0067] At each time step t:

[0068] The actor network μ(s t | θ μ ) selects an action a t , adds Gaussian noise ∈ (∈ ~ N(0, σ 2 )) to get a t = μ(s t | θ μ );

[0069] Execute the action to get the reward r t and the next state s t+1 , and store (s t , a t , r t , s t+1 ) in the experience replay buffer;

[0070] Step 5: Sample from the experience replay buffer and update the network parameters

[0071] Randomly sample a batch of N experience data from the buffer. Randomly sample a batch of N experience data (s i , a i , r i , s i+1 ) from the buffer;

[0072] Update the critic network:

[0073] The target Q-value y i = r i + γQ′(s i+1 , μ′(s i+1 | θ μ′ | θ Q′ );

[0074] The loss function

[0075] Parameter update η Q is the learning rate;

[0076] Update the actor network:

[0077] The loss function

[0078] Parameter update η μ is the learning rate;

[0079] Update the target network:

[0080] Update the parameters θ of the target critic network Q′ ← τθ Q +(1 - τ)θ Q′ ;

[0081] Update the parameters θ of the target actor network μ′ ← τθ μ +(1 - τ)θ μ′ , where τ is the soft update coefficient;

[0082] Continuously repeat the above steps 2 - 4. As the number of interactions between the agent and the environment increases, the actor network and the critic network will gradually learn the optimal turning parameter adjustment strategy, thereby realizing dynamically adjusting the spindle speed, feed rate, and cutting depth of the lathe according to different environmental states to improve the quality and efficiency of turning processing.

[0083] In the S4 chrome plating treatment step, the chrome plating electrolyte formula is: chromic anhydride 250 - 300 g / L, sulfuric acid 2 - 2.5 g / L, additive 0.5 - 1 g / L; the chrome plating temperature is 50 - 55 °C, and the electroplating current density is 30 - 35 A / dm 2 , and the chrome plating time is 2 - 3 hours to form a chrome plating layer with a thickness of 0.045 - 0.055 mm and a hardness of 60 - 62 HRC.

[0084] In the S5 dehydrogenation treatment step, place the mandrel in a heating furnace, heat it to 200 - 250 °C at a heating rate of 50 - 80 °C / h, keep it warm for 3 - 5 hours, and then cool it to room temperature at a cooling rate of 30 - 50 °C / h.

[0085] In the S6 finished product inspection step, use a coordinate measuring instrument to detect that the horizontal length tolerance of the conical angle at the head of the mandrel is within ±0.1 mm, and the conical angle tolerance between 100 mm - 300 mm is within ±0.1 °; use a Rockwell hardness tester to detect the hardness of the chrome plating layer, and the hardness deviation is within ±1 HRC; use a surface roughness instrument to detect that the surface roughness Ra value is not greater than 0.8 μm.

[0086] After the S6 finished product inspection step, adopt a quality traceability and continuous improvement system based on the association rule mining algorithm. This algorithm performs association rule mining on all data in the manufacturing process, and finds out the key factor combinations affecting the quality of the mandrel through the algorithm. According to the mined association rules, establish a quality traceability system. The data mining algorithm established here is as follows:

[0087] Step 1: Data Collection and Preprocessing

[0088] Data collection: Collect all relevant data in the manufacturing process of the mandrel head of the PQF tandem rolling mill, such as blank information (chemical composition, hardness, etc.), processing parameters (temperature, pressure, time, etc.), and finished product inspection results (dimensional accuracy, surface quality, etc.). Organize this data in the form of a transaction database, where each transaction represents a production record.

[0089] Step 2: Generate 1-itemsets and Calculate Support

[0090] Generate 1-itemsets: Extract all individual data items from the transaction database to form the initial candidate itemset C1;

[0091] Calculate support: For each itemset X in C1, calculate its support support(X) using the formula:

[0092]

[0093] where count(X) is the number of transactions containing the itemset X, and N is the total number of transactions in the transaction database;

[0094] Filter frequent 1-itemsets: Set the minimum support threshold min_support, and filter out the itemsets with support greater than or equal to min_support to form the frequent 1-itemset L1;

[0095] Step 3: Generate Candidate Itemsets Layer by Layer and Filter Frequent Itemsets

[0096] For k > 1, repeat the following steps:

[0097] Generate candidate itemset C k : Generate candidate itemset C k-1 through the frequent (k - 1)-itemset L k . The specific method is to connect the itemsets in L k-1 pairwise. If all (k - 1)-subsets of the connected itemset are frequent itemsets, then add it to C k ;

[0098] Calculate support: Scan the transaction database, and for each itemset X in C k , calculate its support support(X) using the formula:

[0099]

[0100] Filter frequent k-itemsets: Set the minimum support threshold min_support, and filter out the itemsets with support greater than or equal to min_support to form the frequent k-itemset L k ;

[0101] Step 4: Generate association rules

[0102] Rule generation: For each frequent item set L, generate all possible non-empty proper subsets X and calculate the confidence of the association rules The confidence is The formula is:

[0103]

[0104] Rule screening: Set the minimum confidence threshold min_confidence and screen out the association rules with confidence greater than or equal to min_confidence;

[0105] Step 5: Establish a quality traceability system

[0106] Determination of key factor combinations: According to the screened association rules, find out the key factor combinations that affect the quality of the mandrel. For example, if the support and confidence of the association rule "billet chemical composition A + processing temperature The mandrel hardness is unqualified" both meet the requirements, then determine "billet chemical composition A + processing temperature B" as the key factor combination affecting the mandrel hardness;

[0107] Construction of the traceability system: Associate these key factor combinations with the production records to establish a quality traceability system. When a quality problem occurs, based on the relevant information of the finished product, trace back to the corresponding production links and key factors to provide a basis for problem investigation and process improvement;

[0108] Step 6: Continuous improvement

[0109] Process adjustment: According to the problems found in the quality traceability system, make targeted adjustments to the manufacturing process. For example, if it is found that a certain key factor combination causes the dimensional accuracy of the mandrel to be unqualified, then adjust the corresponding processing parameters.

[0110] Data update and rule re-mining: As production progresses, continuously update the transaction database, re-run the algorithm to mine association rules, and continuously optimize the quality traceability system and manufacturing process.

[0111] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An improved manufacturing method for the head of a mandrel in a PQF continuous rolling mill, characterized in that, It includes the following steps: S1: Blank screening step Select a mandrel blank suitable for the PQF continuous rolling mill, and conduct a comprehensive inspection on the blank material, size and physical properties; S2: Rough machining step Conduct preliminary turning on the blank to remove surface impurities and surplus, and prepare for subsequent finish turning; S3: Head finish turning step Turn the head of the mandrel to a conical angle horizontal length of 300 mm, and turn a cone with an angle of 12° between 100 mm and 300 mm; S4: Chrome plating treatment step Chrome plate the turned mandrel to make the chrome plating layer have a specific thickness and hardness; S5: Dehydrogenation treatment step Conduct dehydrogenation operation on the chrome-plated mandrel to eliminate internal stress; S6: Finished product inspection step Conduct inspections on multiple indicators including size, hardness and surface quality on the head of the processed mandrel.

2. The improved manufacturing method for the mandrel head of a PQF continuous rolling mill according to claim 1, characterized in that, In the S1 blank screening step, the mandrel blank material is chrome molybdenum alloy steel, in which the chromium content is 2.0% - 2.5%, the molybdenum content is 0.8% - 1.2%, the carbon content is 0.3% - 0.5%, and the rest is iron and inevitable impurities; the blank size tolerance is controlled within ±0.5 mm, and the hardness range is 200 - 220 HBW.

3. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill according to claim 1, characterized in that, In the S2 rough machining step, turning is carried out using a lathe. The lathe spindle speed is 200 - 300 r / min, the feed rate is 0.2 - 0.3 mm / r, and the cutting depth is 1 - 2 mm; the turning tool is selected as a YT15 cemented carbide tool, the tool rake angle is 10° - 15°, and the tool clearance angle is 6° - 8°.

4. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill unit according to claim 1, characterized in that, A surface defect detection system based on machine vision and deep learning algorithm is introduced between the S2 rough machining step and the S3 head finish turning step. The system uses an industrial camera to collect the surface image of the mandrel, and uses a convolutional neural network to perform real-time recognition and classification on scratches, cracks and sand holes defects in the image. Set the defect size threshold. Once a defect exceeding the threshold is detected, automatically adjust the subsequent processing parameters or mark the mandrel.

5. The improved manufacturing method for the mandrel head of a PQF continuous rolling mill according to claim 4, characterized in that, In the S3 head finish turning step, a high-precision CNC lathe is used, the spindle speed is 300 - 400 r / min, the feed rate is 0.05 - 0.1 mm / r, and the cutting depth is 0.1 - 0.2 mm; coolant is used for cooling during the turning process, and the coolant is emulsion with a concentration of 5% - 8%. A surface defect detection system based on machine vision and deep learning algorithm is introduced between the rough machining step and the head finish turning step.

6. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill unit according to claim 1, characterized in that, An adaptive turning parameter optimization algorithm is used in the S3 head finish turning step. This algorithm is based on the deep deterministic policy gradient algorithm of reinforcement learning, and real-time collects environmental state data including tool wear degree, cutting force magnitude and workpiece surface roughness, and dynamically adjusts the spindle speed, feed rate and cutting depth of the lathe.

7. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill according to claim 1, characterized in that, In the S4 chrome plating treatment step, the chrome plating electrolyte formula is: chromic anhydride 250 - 300 g / L, sulfuric acid 2 - 2.5 g / L, additive 0.5 - 1 g / L; the chrome plating temperature is 50 - 55 °C, the electroplating current density is 30 - 35 A / dm2, and the chrome plating time is 2 - 3 hours to form a chrome plating layer with a thickness of 0.045 - 0.055 mm and a hardness of 60 - 62 HRC.

8. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill according to claim 1, characterized in that, In the S5 dehydrogenation treatment step, the mandrel is placed in a heating furnace and heated to 200-250°C at a heating rate of 50-80°C / h, held for 3-5 hours, and then cooled to room temperature at a cooling rate of 30-50°C / h.

9. An improved manufacturing method for the mandrel head of a PQF continuous rolling mill unit according to claim 1, characterized in that, In the S6 finished product inspection step, a coordinate measuring instrument is used to detect that the tolerance of the horizontal length of the conical angle at the head of the mandrel is within ±0.1 mm, and the tolerance of the conical angle is within ±0.1° between a distance of 100 mm and 300 mm; a Rockwell hardness tester is used to detect the hardness of the chromium plating layer, and the hardness deviation is within ±1 HRC; a surface roughness instrument is used to detect that the surface roughness Ra value is not greater than 0.8 μm.

10. An improved manufacturing method for the head of a mandrel of a PQF continuous rolling mill according to claim 1, characterized in that, After the S6 finished product inspection step, a quality traceability and continuous improvement system based on the association rule mining algorithm is adopted. This algorithm performs association rule mining on all data in the manufacturing process, and finds the key factor combinations affecting the quality of the mandrel through the algorithm. According to the mined association rules, a quality traceability system is established.

Images