A seasoning production process parameter regulation method and system

Abstract

The application relates to the technical field of process control, and discloses a seasoning production process parameter control method and system. The method comprises the following steps: resampling and normalizing the temperature and humidity of two material streams in a discharge chute according to the cumulative discharge quality to obtain various sequence quantities under the normalized discharge sequence position, extrapolating the temperature sequence of the material away from the kettle by the difference between the two sequence temperatures at the same sequence position to obtain the off-kettle temperature sequence, taking the point-by-point deviation based on the median of the sequence to form the batch temperature deviation profile, obtaining the batch differentiation characteristics group from the range and the quartile skewness of the profile and verifying the position of the humidity valley and the temperature peak, and mapping the two process threshold limiting correction amounts of the next batch by a fuzzy neural network. The application solves the problems that the position temperature differentiation of the material in the kettle is averaged out by the online detection, the heat dissipation amount of the chute section and the temperature measurement window pollution make the size and direction of the differentiation cannot be determined in the existing seasoning production process parameter control.

Description

Technical Field

[0001] This application relates to the field of process control technology, and in particular to a method and system for controlling process parameters in seasoning production. Background Technology

[0002] The stir-frying process in seasoning production is typically carried out in a horizontal jacketed wok. Heat is provided by jacketed steam, and a scraper paddle stirs and pushes the material inside the wok. After undergoing heating, moisture removal, and flavor compound formation, the material is discharged from the bottom outlet and sent to the granulation or packaging process via a discharge chute. The existing production line uses a formula-driven fixed-value control route for parameter control of this process. Process engineers create fixed process formulas for each variety based on past batch records, and the stir-frying temperature and stirring speed are written into the formula library of the control system. During production, the formula is called according to the variety, and the controller distributes the set values ​​to the jacketed steam regulating valve and the stirring frequency converter, each forming a single-loop closed loop. The detection system is partly located on the wok body, using thermocouples to collect the temperature inside the wok and humidity probes to collect the humidity inside the storage compartment; and partly located on the discharge side, with infrared thermometers and online moisture meters installed on the discharge chute to collect the temperature and moisture content of the discharged material online, and a weighing module to measure the discharge mass. The collected data is sent to the data acquisition and control system via fieldbus to generate trend curves, over-limit alarms, and batch reports. Quality control relies on offline inspection. Moisture content, color difference, and agglomeration rate are measured for each batch, and a sensory evaluation team provides scores. When test results deviate from the standard, process engineers modify the corresponding values ​​in the formula and verify them in the next batch of production.

[0003] The above-mentioned control methods have significant shortcomings. Measuring points on the pot body can only reflect the state of a fixed location within the pot. However, in horizontal woks, materials close to the pot wall are directly heated by the jacket, and materials in the dead zone of the paddle rotation are not stirred for extended periods. Significant temperature variations already exist within the same pot of material, which single-point measurements cannot reflect. While online detection on the discharge side can cover all materials, its output is processed into an average moisture content or a final temperature for that batch. Differences between different parts of the pot are completely erased during the averaging process. The clumping, scorching, and batch-to-batch color differences in the finished seasoning are precisely caused by these differences, not by deviations from the average value. The single-point temperature measurements on the chute contain heat loss from the material within the chute, which cannot be deducted as a fixed value due to variations in workshop ambient temperature and material flow rate. Over long-term use, grease and dust accumulate on the chute's temperature measuring windows. The localized inflated readings caused by window contamination and the material's own regional variations appear identical in the single-channel temperature data. Offline testing involves sampling, submission, and report issuance that spans the entire production cycle of a batch. Test results only apply to subsequent batches; deviations in the current batch are already ingrained in the finished product. Process engineers rely on personal experience to modify formulas, leading to inconsistent correction amounts for the same deviation across different work groups, and the scope of any single modification is unrestricted. Summary of the Invention

[0004] This application provides a method and system for controlling process parameters in seasoning production. This application solves the problems in existing seasoning production process parameter control where the temperature differentiation of different parts of the material in the pot is erased by the average value of online detection, and the heat dissipation of the chute section and contamination of the temperature measuring window make it impossible to determine the magnitude and direction of this differentiation. This application improves upon this by using the discharge sequence as a coordinate to locate the state of different parts of the material and by determining the direction and authenticity of the differentiation.

[0005] In a first aspect, this application provides a method for controlling process parameters in seasoning production, the method comprising: Step S1: Resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence. Step S2: Take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient. Based on the sequence heat dissipation gradient, push the upstream sequence temperature outward towards the chute inlet to obtain the pot exit temperature sequence. Step S3: Using the median of the temperature sequence leaving the pot as a reference, take the point-by-point deviation to obtain the batch temperature deviation profile. Obtain the batch non-uniformity amplitude from the range of the batch temperature deviation profile. Obtain the batch non-uniformity direction from the interquartile skewness of the batch temperature deviation profile. Verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the batch temperature deviation profile using the range of the sequence humidity. The batch non-uniformity amplitude and the batch non-uniformity direction that pass the verification form a batch differentiation feature group. Step S4: Map the intra-batch differentiation feature group to the stirring speed of the current batch using a fuzzy neural network to obtain the intra-batch threshold correction amount, and perform amplitude-limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

[0006] Secondly, this application provides a seasoning production process parameter control system, the seasoning production process parameter control system comprising: The sampling module is used to resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence. The extrapolation module is used to take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient, and extrapolate the upstream sequence temperature towards the chute inlet according to the sequence heat dissipation gradient to obtain the pot exit temperature sequence. The verification module is used to obtain the intra-batch temperature deviation profile by taking the point-by-point deviation based on the median of the outlet temperature sequence, obtain the intra-batch unevenness amplitude from the range of the intra-batch temperature deviation profile, obtain the intra-batch unevenness direction quantity from the interquartile skewness of the intra-batch temperature deviation profile, and verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the intra-batch temperature deviation profile by using the range of the sequence humidity. The intra-batch unevenness amplitude and intra-batch unevenness direction quantity that pass the verification form an intra-batch differentiation feature group. The correction module is used to map the intra-batch differentiation feature group to the stirring speed of the current batch through a fuzzy neural network to obtain the intra-batch threshold correction amount, and to perform a limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

[0007] In the technical solution provided in this application, the three signals in the chute are resampled based on the cumulative discharge mass rather than the sampling time. This ensures that the horizontal axis is determined by the pushing geometry of the scraper blade. Changes in the opening of the discharge butterfly valve, fluctuations in the material flow rate, and pauses during discharge no longer alter the internal position of the boiler corresponding to a certain sequence point. Materials sampled from different batches at the same normalized discharge sequence point point to the same location, and cross-batch comparisons no longer rely on the repeatability of the discharge cycle. The difference between the upstream and downstream sequence temperatures at the same sequence point constitutes the sequence heat dissipation gradient. This difference is composed of the actual temperature drop of the same material within a known travel distance. The effects of workshop ambient temperature and material flow rate on heat dissipation are already included in this difference. The heat dissipation of the chute section is no longer retained in the temperature sequence obtained by extrapolating towards the chute inlet. The drift of ambient temperature during seasonal transitions is not included in the downstream profile calculation. The median of the temperature sequence away from the pot is used as the benchmark for point-by-point deviation. This benchmark is derived from the material itself in this batch, not from a preset value. The proportion of wall-adhering and dead zone materials in the sequence points is less than 20%, so the median is not affected by these two segments. The overall temperature shift caused by batch-to-batch feeding amounts and product switching is eliminated after taking the difference. Give the size of the differentiation, by The differentiation direction is given, with only three quartiles used for quartile skewness. Abnormal readings at a single sequence point do not change the sign. The opposite actions required for the two situations—overheating of the wall layer and undercooking of the central agglomerate—correspond to specific sign intervals. Unidirectional adjustment based on amplitude prevents one situation from being reversed. Verification is performed using the distance between the sequence humidity trough and the sequence position of the positive peak of the profile. Falsely high readings caused by contamination of the temperature measurement window do not have corresponding troughs along the humidity path; such readings are isolated during the verification process and do not enter the threshold correction. Temperature correction limits and rotation speed correction limits are set for the intra-batch threshold correction amount. The correction amplitude of a single batch is constrained within the allowable range of the equipment, and reciprocating adjustments between batches no longer occur.

[0008] In this scheme, the fuzzy neural network does not perform general regression fitting, but rather converts the determined differentiation magnitude and direction into specific correction values ​​for two thresholds in the next batch of formulations. Its adaptability comes from two aspects. First, and The two quantities belong to different dimensions and have different physical meanings. The former is a temperature quantity, and the latter is a dimensionless symbolic quantity. The fuzzification layer takes values ​​according to the Gaussian membership function on its respective universe of discourse. The two do not need to be pre-normalized to the same scale to participate in the multiplication of the same rule. If a multivariate regression with the original values ​​as input is used, the two terms with different dimensions must be manually weighted in advance. Once the weights are fixed, they cannot change with the working point. Secondly, the response of the stir-frying process to the temperature threshold adjustment is not the same under different stirring speeds. The stirring speed of this batch participates in the multiplication of the rule as the third input, so that the same component differentiation feature falls on different rule combinations under different speeds. The first-order form of the consequent makes the correction value change continuously with the differentiation size without jumping. If a lookup table matching or piecewise constant value processing method is used, the correction value will jump at the boundary of the speed when the speed is between two levels. This jump will directly reflect the sudden change of the threshold between two adjacent batches. The membership function center and width of the fuzzification layer are set according to the actual domain of each input quantity. The coefficients of the rule consequent are obtained by training with the differentiation characteristics of historical batches and the paired samples of the agglomeration rate and color difference of the finished products of that batch. The training samples are of the same origin as the observations of this scheme. The mapping learned by the network is bound to the observation of the discharge sequence profile. If the training is replaced with general process parameter samples that are unrelated to this observation, the mapping will lose its basis. Attached Figure Description

[0009] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of an embodiment of the seasoning production process parameter control method in this application; Figure 2 This is a schematic diagram of the temperature deviation from the profile in an embodiment of this application. Figure 3 This is a schematic diagram illustrating the variation of intra-batch non-uniformity amplitude and intra-batch non-uniformity direction with batch in the embodiments of this application; Figure 4 This is a schematic diagram illustrating the verification of the positional consistency between the humidity valley and the temperature peak in the embodiments of this application. Detailed Implementation

[0011] This application provides a method and system for controlling process parameters in the production of seasonings. The terms "first," "second," "third," "fourth," etc. (if present)," in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" or "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or apparatuses.

[0012] For ease of understanding, the specific process of the embodiments of this application is described below. Please refer to [link / reference]. Figure 1 One embodiment of the seasoning production process parameter control method in this application includes: Step S1: Resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence.

[0013] This step deals with solid compound seasonings cooked in a horizontal scraper-jacketed wok. A closed discharge chute is connected to the side outlet at the bottom of the wok, and the scraper blade pushes the material from the wok wall to the discharge outlet. Differences in the location of the material inside the wok cannot be observed by placing monitoring points on the wok body. Hygiene requirements and cleaning operations do not allow for the addition of extended measuring points inside the wok, and the abrasion caused by the blade rotation makes such measuring points unsustainable in the long term. The discharge process itself delivers the material inside the wok in portions according to the pushing order of the scraper blade. The material closest to the wok wall is directly pushed out by the last scraper and leaves the wok first, followed by the main material in the middle of the wok, and the material in the blade rotation dead zone is discharged last. This order is determined by the geometric relationship between the blade and the wok wall. This step treats the discharge process as a step-by-step presentation of the internal location of the wok, collecting temperature and humidity data on the chute, so that the differences in the location of the wok inside are presented in a sequence according to the discharge order.

[0014] Existing methods also incorporate temperature sensors and online moisture meters in the chute, but their records use the sampling time as the x-axis and summarize the entire record as an average moisture content or a final temperature for that batch. Records using sampling time as the x-axis cannot be compared between batches. The opening of the discharge butterfly valve, the instantaneous flow rate, and the pauses caused by bridging during discharge vary from batch to batch. The material corresponding to the same sampling time in different batches does not originate from the same part of the boiler. This step changes this to using the cumulative discharge mass as the x-axis. The cumulative discharge mass corresponds one-to-one with the proportion of material pushed out by the scraper, independent of valve opening and flow rate. The internal position of the boiler corresponding to a certain cumulative discharge mass remains consistent across batches. Then, the cumulative discharge mass is normalized using the total discharge mass of the batch to obtain a normalized discharge sequence with a value ranging from zero to one. The scaling of the horizontal axis caused by the varying amounts of material fed in each batch is thus eliminated.

[0015] In some embodiments, step S1 includes: synchronizing the upstream material flow temperature, downstream material flow temperature, material flow humidity, cumulative discharge mass, and discharge valve position in the discharge chute using a unified timestamp to obtain a batch-wide synchronous sampling point set; performing instantaneous discharge flow rate conversion on the batch-wide synchronous sampling point set based on the ratio of the difference in cumulative discharge mass between two adjacent batch-wide synchronous sampling points to the sampling period to obtain a point-by-point discharge flow rate; and based on the batch-wide synchronous sampling points where the discharge valve position is below the lower limit of the valve position, and the point-by-point discharge flow rate being below the lower limit of the flow rate... Intermittent segments are removed from the batch-wide synchronous sampling points to obtain a continuous discharge sampling point set. Using the cumulative discharge mass as the independent variable and dividing the interval by a set mass step size, the median values ​​of the upstream material flow temperature, downstream material flow temperature, and material flow humidity falling within the same interval in the continuous discharge sampling point set are respectively processed to obtain an order point set. The cumulative discharge mass of the order point set is normalized according to the total discharge mass of this batch to obtain the upstream order temperature, downstream order temperature, and order humidity under the normalized discharge order.

[0016] The sampling period is set to 0.2 seconds, determined by the time it takes for the material flow in the chute to pass through the temperature measurement window. A single batch discharge lasts 300 seconds, with 1500 simultaneous sampling points collected per batch. The lower limit of the valve position is set to 90%, corresponding to the lower edge of the fully open discharge butterfly valve. Below this value, the material flow in the chute exhibits intermittent collapse rather than a continuous columnar shape. The lower limit of the flow rate is set to 1.2 kg / s, calculated from the chute cross-section and the minimum continuous thickness of the material layer. Below this value, the chute wall appears in the field of view of the temperature measurement window instead of the material itself, and the resulting reading does not represent the material temperature. The mass step size is set to 5 kg, ensuring that each sequence point covers approximately three seconds of sampling time and fifteen original sampling points. Using the median within the interval eliminates jumps at single points without smoothing out the true differences between parts. The total output mass of this batch is 498.6 kg, and the single batch feed amount is 500 kg. The difference between the two is the amount of residue adhering to the pot wall and the blade.

[0017] Processing this batch of data using the above parameters, out of 1500 synchronous sampling points, 82 points had valve positions below 90% and flow rates below 1.2 kg / s. After removing these, 1418 points remained, forming a continuous discharge sampling point set. Dividing the data into intervals with a step size of 5 kg, 99 sequence points were obtained. The cumulative discharge mass of sequence point 6 was 30 kg. The normalized discharge sequence... The value is 0.060, representing the upstream order temperature obtained from the median of the interval at that point. 129.65 degrees Celsius, downstream sequence temperature 123.40 degrees Celsius, ordinal humidity The percentage was 2.35%. The cumulative output mass of sequence site 50 was 250 kg. The value is 0.501. 118.40 degrees Celsius The temperature was 112.60 degrees Celsius. The cumulative output mass of sequence site 92 was 460 kg. The value is 0.923. 114.95 degrees Celsius The temperature is 110.30 degrees Celsius.

[0018] Step S2: Take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient. Extrapolate the upstream sequence temperature towards the chute inlet according to the sequence heat dissipation gradient to obtain the boiler discharge temperature sequence.

[0019] Although the sequence measurements correspond to the internal positions of the pot, the material in the chute continuously dissipates heat to the environment during its movement. The temperature readings at each sequence point contain the heat dissipation from the chute section. This heat dissipation is proportional to the temperature difference between the material and the environment. Since the ambient temperature varies with the season and day / night cycle in the workshop, and the material flow rate determines the residence time of the material in the chute, this heat dissipation varies from batch to batch, and the fixed compensation values ​​used in existing methods cannot be changed accordingly. This step involves taking two temperature measurement points on the chute. The same batch of material passes through the upstream measurement point first, then the downstream measurement point. The difference between the two readings represents the actual temperature drop lost by the material during its journey between the two measurement points. The effects of ambient temperature and material flow rate on heat dissipation are already fully included in this difference, eliminating the need to measure the ambient temperature separately. This difference is then divided by the distance between the two measurement points. The heat dissipation gradient is obtained by multiplying it by the extrapolated length of the section from the upstream measuring point to the chute inlet section. The upstream sequence temperature is then added back to obtain the temperature of the material at the instant it leaves the pot, which is recorded as the pot exit temperature. .

[0020] Extrapolation is treated as a first-order linear relationship, which holds true when the heat dissipation gradient within the extrapolated section is equivalent to the heat dissipation gradient between the two measuring points. The chute from the inlet to the downstream measuring point is a closed section with a uniform cross-section; the temperature drop of the material within this section decreases exponentially. The deviation introduced by replacing the exponential relationship with a linear one is... and The ratio increases as the ratio increases, therefore this ratio must be constrained. Existing practices only set up one measuring point on the chute, making it impossible to obtain the heat dissipation gradient, so a fixed value can only be deducted or no deduction is made at all; this step obtains the heat dissipation gradient of this batch and this sequence on the spot by the difference in readings of two measuring points, and the deduction amount varies with the batch and sequence.

[0021] In some embodiments, step S2 includes: pairing the upstream and downstream sequence temperatures at the same normalized discharge sequence position to obtain a sequence temperature pairing value; calculating the quotient of the difference between the pairing sequence temperature values ​​and the center distance of the temperature measurement windows of the upstream and downstream sequence temperatures to obtain a sequence heat dissipation gradient; performing a first-order linear extrapolation increment operation along the chute inlet direction on the sequence heat dissipation gradient based on the extrapolation length from the center of the temperature measurement window of the upstream sequence temperature to the chute inlet section to obtain a sequence extrapolated temperature rise; and performing a superposition process for each of the normalized discharge sequences based on the upstream sequence temperature and the sequence extrapolated temperature rise to obtain a boiler discharge temperature sequence.

[0022] The sequence temperature pairing is completed by the cumulative discharge mass. The upstream and downstream readings at the same cumulative discharge mass correspond to the same batch of material, and the pairing does not depend on the travel time of the material flow in the chute. The sequence heat dissipation gradient is measured in degrees Celsius per meter, and the sequence extrapolated temperature rise is measured in degrees Celsius. The temperature sequence of leaving the pot obtained by superimposing the two with the upstream sequence temperature is calculated point by point according to the following relationship.

[0023]

[0024] In the above formula, For the first The temperature of each sequence point away from the pot, in degrees Celsius; The upstream sequence temperature of this sequence site, in degrees Celsius; This is the downstream sequence temperature of this sequence site, in degrees Celsius. The extrapolated length of the section from the center of the temperature measurement window at the upstream measuring point to the inlet section of the chute is expressed in meters. The distance between the center-to-center temperature measurement windows of the upstream and downstream measuring points is in meters. The temperature difference in the second term on the right side of the equation is in degrees Celsius, the ratio of lengths is dimensionless, and the unit of the sum of the two terms is the same as that on the left side of the equation.

[0025] Take 1.2 meters. This value is determined by the available straight section length of the chute. The two measuring points must be placed in the same cross-sectional section to ensure consistent heat dissipation conditions within the stroke. Take 0.3 meters. The upstream measuring point must avoid the material drop zone at the chute inlet. 0.3 meters is the distance from the end of the drop zone to the starting point of the equal cross-section section. and The ratio is 0.25, not exceeding one-third. Under this ratio, the deviation introduced by replacing the exponential relationship with a linear relationship does not exceed 0.12 degrees Celsius, which is less than the minimum resolvable amount of intra-batch temperature deviation from the profile in this scheme.

[0026] Taking the 6th sequence position as an example, substitute the numerical values ​​according to the above formula.

[0027]

[0028] income The temperature was 131.21 degrees Celsius. The same method was used to treat sequence number 50, with a temperature difference of 5.80 degrees Celsius. The temperature difference at sequence number 92 was 119.85 degrees Celsius; the temperature difference at sequence number 92 was 4.65 degrees Celsius. The temperature is 116.11 degrees Celsius. The 99 values ​​calculated point by point constitute the temperature sequence of the vessel.

[0029] Step S3: Using the median of the temperature sequence leaving the pot as a reference, take the point-by-point deviation to obtain the batch temperature deviation profile. Obtain the batch non-uniformity amplitude from the range of the batch temperature deviation profile. Obtain the batch non-uniformity direction from the interquartile skewness of the batch temperature deviation profile. Verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the batch temperature deviation profile using the range of the sequence humidity. The batch non-uniformity amplitude and the batch non-uniformity direction that pass the verification form a batch differentiation feature group.

[0030] The absolute values ​​of the temperature series after cooking contain the overall temperature level of the batch, which varies with the variety, amount of feed, and cooking time. This method aims to determine the differences between different parts within the same batch of feed; therefore, the overall level must be subtracted first. The benchmark used for subtraction is the median of the batch's own temperature series after cooking. Instead of using preset values, the baseline is provided by the current batch of materials. The overall shift caused by changes in material type and feed quantity is eliminated after taking the difference. The reason for using the median instead of the arithmetic mean is that the wall-adhering layer and dead zone sections account for less than 20% of the 99 sequence points. The arithmetic mean would be affected by the higher and lower values ​​of these two segments. The median represents the temperature level of the main material section. The sequence obtained by taking the difference point by point represents the batch temperature deviation from the profile. .

[0031] In some embodiments, step S3 includes: sorting the temperature values ​​at each normalized discharge sequence position in the discharge temperature sequence in ascending order to obtain a discharge temperature ascending sequence; taking the median value based on the arithmetic mean of the middle item when the sequence length of the discharge temperature ascending sequence is odd and the middle two items when the sequence length is even, to obtain the discharge temperature median; performing a difference operation on each normalized discharge sequence position of the discharge temperature sequence according to the discharge temperature median to obtain the batch temperature deviation profile; extracting the positive maximum value and the negative minimum value of the batch temperature deviation profile to obtain the positive peak value of the profile, the normalized discharge sequence position where the positive peak value of the profile is located, and the negative valley value of the profile, and performing a difference operation on the positive peak value of the profile and the negative valley value of the profile to obtain the batch unevenness amplitude value.

[0032] The number of ordinal positions is 99, which is an odd number. The median is the 50th item in the ascending sequence. Intra-batch uneven amplitude. The difference between the positive peak value and the negative valley value of the profile is used to determine the value, both of which are single-point values, and the quantity is measured in degrees Celsius. The normalized discharge sequence number where the positive peak value of the profile is located is extracted in this step. This sequence number is directly referenced in the subsequent position consistency verification and segment assignment determination, and is not extracted separately.

[0033] The 50th item in the ascending sequence of temperatures leaving the boiler in this batch is 118.90 degrees Celsius. Take this value. The 6th sequence locus. Subtracting 118.90 degrees Celsius from 131.21 degrees Celsius equals 12.31 degrees Celsius; at sequence number 50... The temperature is 0.95 degrees Celsius; at sequence number 92. The temperature was -2.79 degrees Celsius. Of all 99 sequence sites, the positive peak value of the profile was 12.31 degrees Celsius, located at... The value is 0.060; the negative valley value of the profile is −2.79 degrees Celsius. The difference between the two is 15.10 degrees Celsius.

[0034] See Figure 2 The figure shows the intra-batch temperature deviation profiles for batches 1 and 6. The profile for batch 1 shows a significant positive envelope in the wall-adhering layer below the normalized discharge sequence number 0.15, with the envelope apex at 0.060. The profile in the main material section approaches zero and slowly declines, while the profile in the dead material section turns negative. The profile for batch 6 still shows a positive envelope at the same location, but the envelope height decreases to slightly more than one-third of that of batch 1, and the negative value segment in the dead material section also narrows. From the morphology of the two profiles, it can be seen that the temperature differentiation within the pot is concentrated in the wall-adhering layer rather than diffused across all sequences, and this differentiation is gradually compressed after inter-batch correction, rather than occurring abruptly at a single sequence number.

[0035] The range only indicates the magnitude of the differentiation, not its direction. In cases of overheated adhering layers, a small amount of material has a significantly higher temperature while most material remains at a normal level, resulting in a long tail at the upper end of the profile. Conversely, in cases of undercooked central clumps, a small amount of material has a significantly lower temperature while most material remains normal, resulting in a long tail at the lower end of the profile. The ranges for both scenarios can be the same, but the required adjustments are opposite. The former requires lowering the roasting temperature threshold and raising the stirring speed threshold to accelerate the turnover of adhering material, while the latter requires raising the roasting temperature threshold to compensate for the heat input of the main material. Therefore, a separate quantity is needed to distinguish the direction of the long tail. This step uses the interquartile skewness, which consists of the upper quartile, median, and lower quartile values ​​of the profile. Its sign corresponds to the direction of the long tail, and using only three quartiles ensures that abnormal readings at a single sequence point do not change sign. Existing methods use deviation amplitude to unidirectionally correct the formulation, failing to distinguish between these two opposite scenarios.

[0036] In some embodiments, step S3 includes: processing the batch temperature deviation profile by taking the lower quartile, median, and upper quartile values ​​to obtain the lower quartile value, median value, and upper quartile value of the profile; performing quartile skewness calculation based on the quotient obtained by subtracting twice the median value from the sum of the upper quartile value and the lower quartile value of the profile, and dividing the quotient by the difference between the upper quartile value and the lower quartile value of the profile, to obtain the batch non-uniformity direction quantity; and performing minimum value extraction processing on the ordinal humidity to obtain the ordinal humidity. The normalized discharge sequence is determined by the trough value of humidity and the normalized discharge sequence where the trough value of humidity is located. The sequence spacing is calculated by the normalized discharge sequence where the trough value of humidity is located and the normalized discharge sequence where the positive peak value of the profile is located, to obtain the peak-trough sequence spacing. The batch temperature deviation from the profile is verified based on the peak-trough sequence spacing not exceeding the upper limit of the sequence spacing, to obtain the verification pass mark. The batch unevenness amplitude and the batch unevenness direction are combined based on the verification pass mark to obtain the batch differentiation feature group.

[0037] Intra-batch non-uniformity directional quantity This is derived from the relationship shown in the following formula.

[0038]

[0039] In the above formula, The non-uniformity within a batch is a dimensionless quantity. The lower quartile value of the profile, in degrees Celsius; This is the median value of the profile, in degrees Celsius. The values ​​represent the interquartile range on the profile, in degrees Celsius. The numerator is a linear combination of three temperature quantities, in degrees Celsius, and the denominator is the difference between the two temperature quantities, also in degrees Celsius. Dividing the two results in dimensionless division. This is a dimensionless quantity. Because the temperature deviation from the profile within the batch has already been calculated using the median as a reference, If the value is zero, the numerator of the above equation degenerates into the sum of the upper and lower quartile values.

[0040] The direction quantity determination bandwidth is ±0.10. This value is determined by the direction quantity dispersion range of normal batches in this production line. The direction quantity of normal batches falls within this bandwidth. A value greater than 0.10 indicates that the overheating of the wall-mounted layer is the primary cause. When the value is less than -0.10, it is judged that the central cluster is immature and dominant. If the distance falls within the bandwidth, it is considered as a differentiation equilibrium. The upper limit of the sequence spacing is 0.020. This value corresponds to two sequence step sizes. The humidity measurement point and the temperature measurement point are separated by one sequence step size on the chute. The upper limit is obtained by adding a quantization margin of one step size.

[0041] The three quantiles of the temperature deviation from the profile within this batch are: Take -0.80 degrees Celsius. Take 0.00 degrees Celsius. Take 1.40 degrees Celsius. Substitute the values ​​into the formula above.

[0042]

[0043] income The value is 0.2727, which is greater than 0.10. The maximum sequential humidity value in this batch is 3.85%, and the minimum value is 2.10%. The lowest sequential humidity value is located in... At a value of 0.050, the positive peak of the profile is located at 0.060, and the peak-valley sequence interval is... If the value is 0.010 and does not exceed 0.020, the verification pass flag is set to pass.

[0044] In some embodiments, step S3 includes: performing threshold comparison processing on the peak-valley sequence spacing according to the upper limit of the sequence spacing to obtain a verification pass identifier for the pass state or the fail state; performing segment assignment determination processing on the normalized discharge sequence where the positive peak of the profile is located based on the wall-adhering layer segment, the main material segment, and the dead zone material segment divided by the normalized discharge sequence to obtain the differentiation position category; performing partition comparison processing on the intra-batch unevenness direction quantity according to the direction quantity determination bandwidth to obtain the differentiation direction category of wall-adhering layer overheating dominance, central agglomerate under-ripe dominance, or balanced differentiation; and combining the intra-batch unevenness amplitude, the intra-batch unevenness direction quantity, the differentiation position category, and the differentiation direction category based on the verification pass identifier in the pass state to obtain the intra-batch differentiation feature group.

[0045] The material adhering to the wall layer inevitably loses excessive water during overheating, resulting in a trough in its sequence humidity within the same sequence segment. Therefore, the true differentiation of the material leaves traces on both the temperature and humidity tracks simultaneously, and these traces are adjacent to each other. The artificially high readings caused by grease and dust adhering to the temperature measuring window only appear on the temperature track; the humidity track does not have a corresponding trough in this sequence segment, thus the peak-to-trough sequence distance far exceeds the upper limit. Such readings are judged as failing during the verification process and are not subject to threshold correction. The segment division is based on the material pushing sequence of the scraper blades. The normalized discharge sequence 0 to 0.15 is the wall-adhering layer segment, 0.15 to 0.85 is the main material segment, and 0.85 to 1.00 is the dead zone material segment. The boundaries of the three segments are calculated by converting the ratio of the blade sweeping area to the internal volume of the pot cavity.

[0046] The positive peak of the profile in this batch is located at At a value of 0.060, it falls within the wall-attached layer segment, and the differentiation location category is determined to be a wall-attached layer. The value is 0.2727, which is greater than the upper limit of the directional quantity determination bandwidth of 0.10. Therefore, the differentiation direction category is determined to be dominated by overheating of the wall-attached layer. The verification pass indicator is in a passed state. Take 15.10 degrees Celsius. The value of 0.2727 is combined with the two categories mentioned above to form the intra-batch differentiation characteristic group for this batch.

[0047] See Figure 4 This figure shows the results of comparing the positions of humidity troughs and temperature peaks for fifteen consecutive batches. Hollow circles are densely distributed within the diagonal band, with their temperature peaks all falling below 0.075, differing from the humidity troughs by no more than one sequence step. Solid triangles are scattered outside the diagonal band in the lower right, with their temperature peaks located at 0.205, 0.318, and 0.412, respectively, already within the main material section, while the corresponding humidity troughs remain between 0.050 and 0.070. The distribution patterns of these two types of points indicate that the true wall-adhesive layer differentiation follows the positions of both temperature and humidity, while the peaks caused by temperature window contamination appear randomly within the main material section and are unrelated to the humidity troughs. These two types of cases are separate and do not overlap under this coordinate system.

[0048] Step S4: Map the intra-batch differentiation feature group to the stirring speed of the current batch using a fuzzy neural network to obtain the intra-batch threshold correction amount, and perform amplitude-limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

[0049] After determining the magnitude, direction, and authenticity of the differentiation, it is necessary to convert it into specific correction values ​​for the two thresholds in the next batch of formula. This conversion is not a simple proportional relationship. For temperature measurement Both are dimensionless symbolic quantities, and their effects on the correction value are coupled and the same. Different The corresponding correction directions are opposite; the stirring speed of this batch determines the response range of the frying process to the temperature threshold adjustment, and the required correction values ​​for the same component's chemical characteristics are different at low and high stirring speeds. This step uses a fuzzy neural network for this conversion. The network has three input layers, nine fuzzification layers, twenty-seven regularization layers, twenty-seven normalization layers, and two output layers. The inputs are as follows: , With the stirring speed of this batch The output is the correction amount for the frying temperature threshold. Correction amount for stirring speed threshold .

[0050] In some embodiments, step S4 includes: inputting the intra-batch unevenness amplitude, the intra-batch unevenness direction, and the batch stirring speed from the intra-batch differentiation feature group into the fuzzification layer of the fuzzy neural network; performing membership degree processing according to the Gaussian membership degree function set on the domain of each input quantity to obtain an input membership degree set; performing rule-by-rule multiplication on the membership degrees belonging to different input quantities in the input membership degree set to obtain the rule activation intensity; and normalizing each rule activation intensity according to the sum of all the rule activation intensities to obtain a normalized value. Activation intensity; The first-order consequents of each rule are processed according to the intra-batch non-uniformity amplitude and the intra-batch non-uniformity direction to obtain the rule consequent value, and the rule consequent value is weighted and summed based on the normalized activation intensity to obtain the intra-batch threshold correction amount; The intra-batch threshold correction amount is truncated according to the temperature correction limit value and the speed correction limit value to obtain the limited correction amount, and the roasting temperature threshold and the stirring speed threshold of the current batch are superimposed based on the limited correction amount to obtain the roasting temperature threshold and the stirring speed threshold of the next batch.

[0051] The membership degree of the fuzzy layer is obtained by the relationship shown in the following formula.

[0052]

[0053] In the above formula, For the first Item input quantity in the first Membership degree on a membership function, dimensionless; For the first Item input quantity, When 1 is taken The unit is Celsius, and 2 is taken as 2 degrees Celsius. Dimensionless, take 3 as The unit is revolutions per minute; The center of this membership function is denoted by and its unit is . Consistent; The width of the membership function, its unit is . Consistent. The numerator and denominator of the exponent have the same unit, and the argument is a dimensionless quantity.

[0054] The domain of discussion is set to 0 to 30 degrees Celsius, with the three centers set to 5, 15 and 25 degrees Celsius respectively, and the width of each center set to 5 degrees Celsius. The upper limit of this domain of discussion is determined by the maximum temperature difference between the wall layer of the wok and the main material in this production line. The universe of discourse ranges from -1 to 1, with the three centers at -0.5, 0, and 0.5 respectively, and the width of each is 0.4. This universe of discourse represents the inherent range of values ​​for the interquartile skewness. The universe of discourse is set at 20 to 60 revolutions per minute, with the three centers set at 25, 40, and 55 revolutions per minute respectively, and the width of each center set at 7.5 revolutions per minute. This universe of discourse is determined by the speed range of the mixing inverter. Each rule in the rule layer is obtained by multiplying the membership degrees of the three inputs together. Each of the three inputs has three membership functions, resulting in a total of twenty-seven rules.

[0055] This batch Using 15.10 degrees Celsius, the three membership degrees are 0.0169, 0.9996, and 0.0198, respectively. This batch... Take 0.2727, and substitute its value on the membership function with center 0.5 into the above formula.

[0056]

[0057] Calculated using the same method The values ​​for the other two membership functions are 0.0240 and 0.6283. This batch With a revolutions per minute of 42, the three membership degrees are 0.0059, 0.9314, and 0.0495, respectively.

[0058] Rule activation strength The result is the dimensionless product of the three membership degrees taken by this rule. Since the activation strengths of each rule vary, they must be normalized before being used for weighting. The normalized activation strengths are... This is derived from the relationship shown in the following formula.

[0059]

[0060] In the above formula, For the first The normalized activation intensity of the rule is dimensionless; For the first The activation strength of a rule is obtained by multiplying the three membership degrees acquired by that rule, and is dimensionless; the summation iterates through all twenty-seven rules. Both the numerator and denominator are dimensionless quantities. Dimensionless, and all The sum of these terms equals one, and this term must be verified in the calculation example.

[0061] Rule number 2 is the membership function corresponding to the center of each of the three inputs. Centered Take the center value as 0.5. The rule corresponding to centering is Rule 1. Rule 1... Multiplying 0.9996, 0.7240, and 0.9314 together yields 0.6741; (Item 2) Multiplying 0.9996, 0.6283, and 0.9314 together yields 0.5850. Since the activation strength is the product of three terms, the sum of the activation strengths of the twenty-seven rules equals the product of the sums of the membership degrees of the three inputs. These sums are 1.0363, 1.3763, and 0.9868, respectively, and their product is 1.4074. The normalized activation strength of the first rule is then substituted into the above formula.

[0062]

[0063] The second rule can be calculated using the same method. The sum of the normalized activation strengths of the remaining twenty-five rules is 0.4157, and the sum of the normalized activation strengths of the other twenty-five rules is 0.1053. The sum of the three equals 1.0000, and both nonnegativity and normalization hold true.

[0064] The consequent of the rule takes a first-order form, meaning it is a linear combination of two inputs. This form ensures that the correction value changes continuously with the magnitude of differentiation without abruptly changing at the rule boundaries. (Treatment amount for frying temperature threshold) This is derived from the relationship shown in the following formula.

[0065]

[0066] In the above formula, This is the correction amount for the stir-frying temperature threshold, in degrees Celsius. For the first The normalized activation intensity of the rule is dimensionless; This is the coefficient of the batch non-uniform amplitude term in the subsequent clause of this rule, and it is dimensionless. This is the coefficient of the non-uniformity directional quantity term within a batch in the subsequent part of this rule, in degrees Celsius; This is a constant term for the consequent of this rule, in degrees Celsius. The three terms within parentheses are all in degrees Celsius, consistent with dimensionless units. The product and summation result is still in degrees Celsius, consistent with the left side of the equals sign. Stirring speed threshold correction amount. The three coefficients are derived from the consequent form of the same structure and are denoted as follows: , and , The unit is revolutions per minute per degree Celsius. and The unit is revolutions per minute, and the income is... The unit is revolutions per minute (RPM). Each coefficient is obtained through backpropagation training using paired samples of the intra-batch differentiation characteristic groups from historical batches and the agglomeration rate and color difference of the finished products in that batch. The learning rate is 0.01, and the iteration termination condition is that the change in mean square error over twenty consecutive iterations is less than [a certain value]. .

[0067] The consequent coefficient of rule 1 is, Take −0.32, Take -6.0 degrees Celsius. Taking 0 degrees Celsius, the consequent value is -0.32 multiplied by 15.10 plus -6.0 multiplied by 0.2727, which equals -6.468 degrees Celsius; Rule 2 Take -0.16, Take -4.0 degrees Celsius. Taking 0 degrees Celsius, its consequent value is -3.507 degrees Celsius; the consequent values ​​of the remaining twenty-five rules, after being weighted by their normalized activation intensities, total -0.212 degrees Celsius. Substitute the values ​​into the formula above.

[0068]

[0069] income It is −4.77 degrees Celsius. The same method was used to calculate the speed side, according to rule number 1. Take 0.24 revolutions per minute per degree Celsius. Taking 4.0 revolutions per minute, the consequent is 4.715 revolutions per minute; the consequent of rule 2 is 2.466 revolutions per minute; the weighted sum of the remaining rules is 0.148 revolutions per minute; the result is... It is 3.43 revolutions per minute.

[0070] The temperature correction limit is ±5.0 degrees Celsius, and the speed correction limit is ±5.0 revolutions per minute. Both limits are determined by the single-step setpoint increment that the jacketed steam regulating valve and the stirring inverter can withstand. Steps exceeding these limits will cause overshoot in the heating phase of the next batch. This batch -4.77 degrees Celsius The rpm is 3.43 revolutions per minute, both within the limits, and can be used directly. This batch's roasting temperature threshold... The temperature is 122.0 degrees Celsius, and the stirring speed threshold for this batch is 42 revolutions per minute. After superimposing these values, the stir-frying temperature threshold for the next batch is 117.23 degrees Celsius, and the stirring speed threshold for the next batch is 45.43 revolutions per minute. This information is written into the process formula for the next batch, and the controller calls it up when starting the next batch and sends control commands to the jacket steam regulating valve and the stirring frequency converter.

[0071] Repeat steps S1 to S4 for the next batch of material output. The convergence criterion is set to three consecutive batches. Not exceeding 8.0 degrees Celsius and The absolute value does not exceed 0.10. Once the criterion is met, the write threshold is frozen and the system enters monitoring mode. Writes resume when any batch exceeds the criterion. The actual test results for six consecutive batches are as follows: The temperatures are 15.10, 11.40, 8.90, 7.30, 6.40, and 5.90 degrees Celsius, respectively. The values ​​are 0.2727, 0.1893, 0.1104, 0.0682, 0.0511, and 0.0403, respectively. These values ​​were calculated for each batch. The temperatures are -4.77, -2.88, -1.73, -0.94, -0.61, and -0.42 degrees Celsius, respectively. The stirring speeds were 3.43, 2.29, 1.30, 0.70, 0.45, and 0.31 revolutions per minute, respectively. The stir-frying temperature threshold decreased from 122.0 degrees Celsius to 110.65 degrees Celsius, passing through 117.23, 114.35, 112.62, 111.68, and 111.07 degrees Celsius. The stirring speed threshold increased from 42 revolutions per minute to 50.48 revolutions per minute, passing through 45.43, 47.72, 49.02, 49.72, and 50.17 revolutions per minute. (Batches 4 through 6...) All not exceeding 8.0 degrees Celsius and The absolute value of each value does not exceed 0.10. The criteria are met for three consecutive batches. The threshold is frozen after the sixth batch.

[0072] See Figure 3 This figure illustrates the intra-batch uneven amplitude and intra-batch uneven direction of six consecutive batches. The height of the left-hand column decreases batch by batch from the first batch, with the third batch crossing the convergence criterion value indicated by the horizontal dashed line, and the subsequent three batches remaining below this line. Similarly, the height of the right-hand column decreases batch by batch, with the fourth batch and subsequent batches falling below the criterion value indicated by the horizontal dotted line. The decreasing step size of both sets of columns narrows batch by batch without any rebound, indicating that the inter-batch correction does not produce a cyclical adjustment. The two criteria are met sequentially rather than simultaneously, with the amplitude criterion being met earlier than the direction criterion in one batch.

[0073] The above describes the method for controlling the process parameters of seasoning production in the embodiments of this application. The following describes the system for controlling the process parameters of seasoning production in the embodiments of this application. One embodiment of the system for controlling the process parameters of seasoning production in the embodiments of this application includes: The sampling module is used to resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence. The extrapolation module is used to take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient, and extrapolate the upstream sequence temperature towards the chute inlet according to the sequence heat dissipation gradient to obtain the pot exit temperature sequence. The verification module is used to obtain the intra-batch temperature deviation profile by taking the point-by-point deviation based on the median of the outlet temperature sequence, obtain the intra-batch unevenness amplitude from the range of the intra-batch temperature deviation profile, obtain the intra-batch unevenness direction quantity from the interquartile skewness of the intra-batch temperature deviation profile, and verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the intra-batch temperature deviation profile by using the range of the sequence humidity. The intra-batch unevenness amplitude and intra-batch unevenness direction quantity that pass the verification form an intra-batch differentiation feature group. The correction module is used to map the intra-batch differentiation feature group to the stirring speed of the current batch through a fuzzy neural network to obtain the intra-batch threshold correction amount, and to perform a limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

[0074] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for controlling process parameters in seasoning production, characterized in that, The method includes: Step S1: Resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence. Step S2: Take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient. Based on the sequence heat dissipation gradient, push the upstream sequence temperature outward towards the chute inlet to obtain the pot exit temperature sequence. Step S3: Using the median of the temperature sequence leaving the pot as a reference, take the point-by-point deviation to obtain the batch temperature deviation profile. Obtain the batch non-uniformity amplitude from the range of the batch temperature deviation profile. Obtain the batch non-uniformity direction from the interquartile skewness of the batch temperature deviation profile. Verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the batch temperature deviation profile using the range of the sequence humidity. The batch non-uniformity amplitude and the batch non-uniformity direction that pass the verification form a batch differentiation feature group. Step S4: Map the intra-batch differentiation feature group to the stirring speed of the current batch using a fuzzy neural network to obtain the intra-batch threshold correction amount, and perform amplitude-limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

2. The method for controlling the process parameters of seasoning production according to claim 1, characterized in that, Step S1 includes: The upstream material flow temperature, downstream material flow temperature, material flow humidity, cumulative discharge mass, and discharge valve position in the discharge chute are synchronized and aligned according to a unified timestamp to obtain a batch of synchronized sampling points. The instantaneous discharge flow rate of the batch synchronous sampling point set is converted by the ratio of the cumulative discharge mass difference between two adjacent batch synchronous sampling points to the sampling period, so as to obtain the point-by-point discharge flow rate. Based on the batch synchronous sampling points where the discharge valve position is lower than the lower limit and the batch synchronous sampling points where the discharge flow rate is lower than the lower limit, discontinuous segment rejection processing is performed to obtain a continuous discharge sampling point set. Using the cumulative discharge mass as the independent variable and dividing the interval by a set mass step size, the median values ​​of the upstream material flow temperature, the downstream material flow temperature, and the material flow humidity of the continuous discharge sampling points falling into the same interval are respectively processed to obtain the sequence point set. Then, the cumulative discharge mass of the sequence point set is normalized according to the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature, and sequence humidity under the normalized discharge sequence.

3. The method for controlling the process parameters of seasoning production according to claim 2, characterized in that, Step S2 includes: The upstream and downstream sequence temperatures at the same normalized discharge sequence position are paired and processed to obtain the sequence temperature pairing value. The sequential heat dissipation gradient is obtained by dividing the difference between the temperature measurement window center distances of the upstream and downstream sequential temperatures by the quotient of the temperature measurement window center distances. Based on the extrapolation length of the temperature measurement window center of the upstream sequence temperature to the chute inlet section, the first-order linear extrapolation increment calculation is performed on the sequence heat dissipation gradient along the chute inlet direction to obtain the sequence extrapolation temperature rise. Based on the upstream sequence temperature and the sequence extrapolated temperature rise, the normalized discharge sequence is superimposed one by one to obtain the discharge temperature sequence.

4. The method for controlling the process parameters of seasoning production according to claim 3, characterized in that, The step of taking point-by-point deviations based on the median of the temperature sequence leaving the pot to obtain the batch temperature deviation profile, and obtaining the batch non-uniformity amplitude from the range of the batch temperature deviation profile, includes: The temperature values ​​at each normalized discharge sequence position in the discharge temperature sequence are sorted in ascending order to obtain the discharge temperature ascending sequence. The median of the temperature at which the pot is removed is obtained by taking the middle term when the sequence length is odd and the arithmetic mean of the two middle terms when the sequence length is even. The difference between the normalized discharge sequence positions is calculated by performing a stepwise difference operation on the discharge temperature sequence based on the median discharge temperature. This yields the batch temperature deviation profile. The maximum positive value and minimum negative value of the temperature deviation from the contour within the batch are extracted to obtain the positive peak value of the contour, the normalized discharge sequence position where the positive peak value of the contour is located, and the negative valley value of the contour. The difference between the positive peak value of the contour and the negative valley value of the contour is calculated to obtain the amplitude of the non-uniformity within the batch.

5. The method for controlling the process parameters of seasoning production according to claim 4, characterized in that, The intra-batch non-uniformity direction quantity is obtained from the interquartile skewness of the intra-batch temperature deviation profile, and the positional consistency between the valley position of the normalized discharge sequence and the peak position of the intra-batch temperature deviation profile is verified by the range of the sequence humidity. The intra-batch non-uniformity amplitude and the intra-batch non-uniformity direction quantity that pass the verification are combined to form an intra-batch differentiation characteristic group, including: The temperature deviation from the contour within the batch is processed by taking the lower quartile, median, and upper quartile values ​​to obtain the lower quartile, median, and upper quartile values ​​of the contour. The quartile skewness is calculated by subtracting twice the median value of the contour from the sum of the upper quartile value and the lower quartile value of the contour, and then dividing the quotient by the difference between the upper quartile value and the lower quartile value of the contour. This yields the intra-batch non-uniformity direction quantity. The minimum value extraction process is performed on the sequence humidity to obtain the sequence humidity valley value and the normalized discharge sequence where the sequence humidity valley value is located. Then, the sequence interval calculation process is performed based on the normalized discharge sequence where the sequence humidity valley value is located and the normalized discharge sequence where the profile positive peak value is located to obtain the peak-valley sequence interval. The batch temperature deviation profile is verified based on the peak-valley sequence interval not exceeding the upper limit of the sequence interval. A verification pass mark is obtained, and the batch non-uniformity amplitude and the batch non-uniformity direction are combined according to the verification pass mark to obtain the batch differentiation feature group.

6. The method for controlling the process parameters of seasoning production according to claim 5, characterized in that, The batch temperature deviation profile verification process, based on the peak-valley sequence interval not exceeding the upper limit of the sequence interval, is performed to obtain a verification pass identifier. Then, the batch non-uniformity amplitude and the batch non-uniformity direction are combined according to the verification pass identifier to obtain a batch differentiation feature group, including: The peak-valley sequence spacing is compared with the upper limit of the sequence spacing to obtain the verification pass mark of the pass state or the fail state; Based on the normalized discharge sequence, the wall-adhering layer segment, the main material segment, and the dead zone segment are divided, and the normalized discharge sequence where the positive peak of the contour is located is processed to determine the segment classification and obtain the differentiation position category. Based on the bandwidth determination of the direction quantity, the non-uniform direction quantity within the batch is partitioned and compared to obtain the differentiation direction category of overheated wall layer, under-mature central agglomerate, or balanced differentiation. Based on the verification pass identifier in the pass state, the intra-batch unevenness amplitude, the intra-batch unevenness direction, the differentiation position category, and the differentiation direction category are combined to obtain the intra-batch differentiation feature group.

7. The method for controlling the process parameters of seasoning production according to claim 6, characterized in that, Step S4 includes: The intra-batch unevenness amplitude, intra-batch unevenness direction, and batch stirring speed in the intra-batch differentiation feature group are respectively input into the fuzzification layer of the fuzzy neural network, and the membership degree is processed according to the Gaussian membership degree function set on the domain of each input quantity to obtain the input membership degree set. The membership degrees belonging to different input quantities in the input membership degree set are processed by a rule-by-rule multiplication operation to obtain the rule activation strength. Then, the activation strength of each rule is normalized according to the sum of all the rule activation strengths to obtain the normalized activation strength. The first-order consequent of each rule is processed by taking values ​​based on the intra-batch non-uniform amplitude and the intra-batch non-uniform direction to obtain the rule consequent value. The rule consequent value is then weighted and summed based on the normalized activation intensity to obtain the intra-batch threshold correction amount. The batch threshold correction amount is truncated based on the temperature correction limit and the speed correction limit to obtain the limited correction amount. The batch frying temperature threshold and the batch stirring speed threshold are then superimposed based on the limited correction amount to obtain the frying temperature threshold and stirring speed threshold for the next batch.

8. A seasoning production process parameter control system, characterized in that, For implementing the seasoning production process parameter control method as described in any one of claims 1-7, the seasoning production process parameter control system comprises: The sampling module is used to resample the upstream material flow temperature, downstream material flow temperature and material flow humidity in the discharge chute according to the cumulative discharge mass, and normalize them with the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature and sequence humidity under the normalized discharge sequence. The extrapolation module is used to take the difference between the upstream sequence temperature and the downstream sequence temperature at the same normalized discharge sequence position to obtain the sequence heat dissipation gradient, and extrapolate the upstream sequence temperature towards the chute inlet according to the sequence heat dissipation gradient to obtain the pot exit temperature sequence. The verification module is used to obtain the intra-batch temperature deviation profile by taking the point-by-point deviation based on the median of the outlet temperature sequence, obtain the intra-batch unevenness amplitude from the range of the intra-batch temperature deviation profile, obtain the intra-batch unevenness direction quantity from the interquartile skewness of the intra-batch temperature deviation profile, and verify the positional consistency between the valley position of the normalized discharge sequence and the peak position of the intra-batch temperature deviation profile by using the range of the sequence humidity. The intra-batch unevenness amplitude and intra-batch unevenness direction quantity that pass the verification form an intra-batch differentiation feature group. The correction module is used to map the intra-batch differentiation feature group to the stirring speed of the current batch through a fuzzy neural network to obtain the intra-batch threshold correction amount, and to perform a limiting correction on the frying temperature threshold and stirring speed threshold of the next batch according to the intra-batch threshold correction amount.

9. The system according to claim 8, characterized in that, The upstream material flow temperature, downstream material flow temperature, and material flow humidity in the discharge chute are resampled according to the cumulative discharge mass, and normalized using the total discharge mass of this batch to obtain the normalized upstream sequence temperature, downstream sequence temperature, and sequence humidity at the discharge sequence, including: The upstream material flow temperature, downstream material flow temperature, material flow humidity, cumulative discharge mass, and discharge valve position in the discharge chute are synchronized and aligned according to a unified timestamp to obtain a batch of synchronized sampling points. The instantaneous discharge flow rate of the batch synchronous sampling point set is converted by the ratio of the cumulative discharge mass difference between two adjacent batch synchronous sampling points to the sampling period, so as to obtain the point-by-point discharge flow rate. Based on the batch synchronous sampling points where the discharge valve position is lower than the lower limit and the batch synchronous sampling points where the discharge flow rate is lower than the lower limit, discontinuous segment rejection processing is performed to obtain a continuous discharge sampling point set. Using the cumulative discharge mass as the independent variable and dividing the interval by a set mass step size, the median values ​​of the upstream material flow temperature, the downstream material flow temperature, and the material flow humidity of the continuous discharge sampling points falling into the same interval are respectively processed to obtain the sequence point set. Then, the cumulative discharge mass of the sequence point set is normalized according to the total discharge mass of this batch to obtain the upstream sequence temperature, downstream sequence temperature, and sequence humidity under the normalized discharge sequence.

10. The system according to claim 9, characterized in that, The difference between the upstream and downstream temperatures at the same normalized discharge sequence position is used to obtain the sequence heat dissipation gradient. The upstream temperature is then extrapolated towards the chute inlet according to this heat dissipation gradient to obtain the boiler discharge temperature sequence, including: The upstream and downstream sequence temperatures at the same normalized discharge sequence position are paired and processed to obtain the sequence temperature pairing value. The sequential heat dissipation gradient is obtained by dividing the difference between the temperature measurement window center distances of the upstream and downstream sequential temperatures by the quotient of the temperature measurement window center distances. Based on the extrapolation length of the temperature measurement window center of the upstream sequence temperature to the chute inlet section, the first-order linear extrapolation increment calculation is performed on the sequence heat dissipation gradient along the chute inlet direction to obtain the sequence extrapolation temperature rise. Based on the upstream sequence temperature and the sequence extrapolated temperature rise, the normalized discharge sequence is superimposed one by one to obtain the discharge temperature sequence.

Images