Device Parameter Adjustment Method and System

Abstract

A device parameter adjustment method and system are provided. The device parameter adjustment method includes: obtaining a training data set regarding the historical operation information of the device, where each piece of training data includes a natural parameter value, a device parameter value, and a device operation state value at a corresponding moment; training a cold start model for predicting the device operation state based on the natural parameter value and the device parameter based on the training data set; obtaining the current natural parameter value and the target operation state value of the device, and determining the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operation state value; selecting the final device parameter value from the candidate device parameter values of the device, and adjusting the device parameter of the device according to the final device parameter value.

Description

Technical Field

[0001] Generally speaking, the present invention relates to the field of artificial intelligence, and more specifically, to a method and system for adjusting device parameters. Background Art

[0002] Currently, the adjustment of device parameter values of a device is generally based on business experience and expert rules. However, the existing adjustment method of device parameter values has the following disadvantages: due to the need for long-term investment in human resources, there are operational risks, and the experience may become invalid over time, and the adjustment experience is not easy to precipitate and inherit. In addition, the existing method for adjusting device parameter values is strongly coupled with the business and does not have the ability to be widely promoted or migrated for use. Summary of the Invention

[0003] The exemplary embodiments of the present invention can at least solve the above problems, or may not solve the above problems.

[0004] In one aspect of the present invention, there is provided a method for adjusting device parameters, where the method includes: obtaining a training data set of historical operation information of the device, where each piece of training data includes a natural parameter value, a device parameter value, and a device operation state value at a corresponding moment; training a cold start model for predicting the device operation state based on the natural parameter value and the device parameter value based on the training data set; obtaining the current natural parameter value and the target operation state value of the device, and determining the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operation state value; selecting the final device parameter value from the candidate device parameter values of the device, and adjusting the device parameter of the device according to the final device parameter value.

[0005] In one implementation, the step of determining the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operation state value includes: determining the number of device parameter values of the device;

[0006] If the number is less than a predetermined threshold, the following processing is performed: combining all possible values of the device parameter values one by one to obtain multiple groups of device parameter values, respectively inputting each group of device parameter values in the multiple groups of device parameter values and the current natural parameter value into the trained cold start model to obtain multiple groups of predicted operation data values, respectively comparing each group of predicted operation data values in the multiple groups of predicted operation data values with the target operation data value, and selecting the multiple groups of device parameter values corresponding to the most matching predicted operation data value as the candidate device parameter values;

[0007] If the number is not less than the predetermined threshold, a heuristic search algorithm is used to calculate and determine the candidate device parameter value of the device.

[0008] In one embodiment, the step of selecting the final device parameter value from the candidate device parameter values of the device includes: when the candidate device parameter values include multiple groups of candidate device parameter values, selecting one group of candidate device parameter values as the device parameter value.

[0009] In one embodiment, the step of selecting one group of candidate device parameter values as the device parameter value includes: performing the following processing for each group of candidate device parameter values: determining a first change amount of the target operation data value when adding a corresponding interference parameter value to each current natural parameter value; determining a second change amount of the target operation data value when adding a corresponding interference parameter value to each candidate parameter value in the group of candidate device parameter values; based on the determined first change amount and second change amount of the target operation data value, determining a stability index of the target operation data value under the group of candidate device parameter values, and taking the candidate device parameter value corresponding to the minimum value among the stability indexes of the target operation data values obtained under the multiple groups of candidate device parameter values as the device parameter value.

[0010] In one embodiment, the step of selecting one group of candidate device parameter values as the device parameter value includes: performing interpolation processing on the multiple groups of candidate device parameter values to obtain the expanded multiple groups of candidate device parameter values; determining a stability index of the target operation data at possible values of the expanded multiple groups of candidate device parameter values; and taking the candidate device parameter value corresponding to the minimum value among the obtained stability indexes as the device parameter value.

[0011] In one embodiment, the method further includes: controlling the device to operate according to the determined device parameter value; collecting in real time the true value of the operation data generated after the device operates; comparing the true value of the operation data with the target operation data value; and adjusting the device parameter value based on the comparison result.

[0012] In one embodiment, the step of adjusting the device parameter value includes: constructing a fine-tuning model for predicting the comparison result between the true value of the operation data and the target operation data value based on the natural parameter value, the device parameter value, and the device parameter adjustment value; obtaining a fine-tuning training data set, where the fine-tuning training data set includes historical natural parameter values, historical device parameter values, historical device parameter adjustment values of the device, and the comparison results between the historical true values of the operation data and the target operation state values; iteratively updating the fine-tuning model based on the fine-tuning training data set; determining the device parameter adjustment value based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the true value of the operation data and the target operation data value; and determining the adjusted device parameter value based on the determined device parameter value and the device parameter adjustment value.

[0013] In one embodiment, the device is a tobacco dryer, the natural parameter value includes a combination of one or more of tobacco moisture, ambient temperature and ambient humidity, the equipment parameter value includes a combination of one or more of hot air damper opening, hot air temperature, cylinder speed, dehumidification damper opening and steam pressure of the rehumidification device, and the equipment operating status value includes the moisture content of the tobacco.

[0014] In one embodiment, the first change in the target operating data value is determined by the following formula:

[0015]

[0016] Wherein, Y is the target operation data value, and the first change of the target operation data value Y is Q w is the parameter value of the wth group of candidate devices, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

[0017] In one embodiment, the second change in the target operating data value is determined by the following formula:

[0018]

[0019] Wherein, Y is the target operation data value, and the second variation of the target operation data value Y is Q w is the parameter value of the wth group of candidate devices, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the w-th group of candidate device parameter values, and λ is the interference parameter value corresponding to the j-th device parameter value.

[0020] In one embodiment, the stability index of the target operating data value under the wth set of candidate device parameter values is determined by the following formula:

[0021]

[0022] Among them, Y is the target running data value, S w is the stability index of the target operating data value under the wth group of candidate equipment parameter values, Δp i is the natural parameter p i Sensitivity, Δq j is the device parameter q j sensitivity.

[0023] Another aspect of the present invention provides a device parameter adjustment system, wherein the system includes a training data acquisition module, a model training module, a candidate parameter determination module, and a device parameter adjustment module; the training data acquisition module is configured to acquire a training data set regarding the historical operation information of the device, wherein each piece of training data includes a natural parameter value, a device parameter value, and a device operation state value at a corresponding moment; the model training module is configured to train a cold start model for predicting the device operation state based on the natural parameter value and the device parameter value based on the training data set; the candidate parameter determination module is configured to acquire the current natural parameter value and the target operation state value regarding the device, and determine the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operation state value; the device parameter adjustment module is configured to select the final device parameter value from the candidate device parameter values of the device, and adjust the device parameter of the device according to the final device parameter value.

[0024] In one embodiment, the candidate parameter determination module is configured to: determine the number of device parameter values regarding the device; if the number is less than a predetermined threshold, perform the following processing: combine all possible values of the device parameter values one by one to obtain multiple groups of device parameter values, input each group of device parameter values in the multiple groups of device parameter values and the current natural parameter value into the trained cold start model respectively to obtain multiple groups of predicted operation data values, compare each group of predicted operation data values in the multiple groups of predicted operation data values with the target operation data value respectively, and select the multiple groups of device parameter values corresponding to the most matching predicted operation data value as the candidate device parameter values; if the number is not less than the predetermined threshold, use a heuristic search algorithm to calculate and determine the candidate device parameter value of the device.

[0025] In one embodiment, the device parameter adjustment module is configured to: when the candidate device parameter values include multiple groups of candidate device parameter values, select one group of candidate device parameter values as the device parameter value.

[0026] In one embodiment, the device parameter adjustment module is configured to: perform the following processing for each group of candidate device parameter values: determine the first change amount of the target operation data value when adding a corresponding interference parameter value to each current natural parameter value; determine the second change amount of the target operation data value when adding a corresponding interference parameter value to each candidate parameter value in the group of candidate device parameter values; based on the determined first change amount and second change amount of the target operation data value, determine the stability index of the target operation data value under the group of candidate device parameter values, and use the candidate device parameter value corresponding to the minimum value of the stability indexes of the target operation data values obtained under the multiple groups of candidate device parameter values as the device parameter value.

[0027] In one embodiment, the device parameter adjustment module is configured to: perform interpolation processing on the multiple groups of candidate device parameter values to obtain the expanded multiple groups of candidate device parameter values; determine the stability index of the target operation data under the possible values of the expanded multiple groups of candidate device parameter values; and use the candidate device parameter value corresponding to the minimum value in the obtained stability indexes as the device parameter value.

[0028] In one embodiment, the system further includes a device operation control module, which is configured to: control the device to operate according to the determined device parameter value; collect in real time the true value of the operation data generated after the device operates; compare the true value of the operation data with the target operation data value; and adjust the device parameter value based on the comparison result.

[0029] In one embodiment, the device operation control module is configured to: construct a fine-tuning model for predicting the comparison result between the true value of the operation data and the target operation data value based on the natural parameter value, the device parameter value, and the device parameter adjustment value; obtain a fine-tuning training data set, where the fine-tuning training data set includes the historical natural parameter values, historical device parameter values, historical device parameter adjustment values of the device, and the comparison results between the historical true values of the operation data and the target operation state values; iteratively update the fine-tuning model based on the fine-tuning training data set; determine the device parameter adjustment value based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the true value of the operation data and the target operation data value; and determine the adjusted device parameter value based on the determined device parameter value and the device parameter adjustment value.

[0030] In one embodiment, the device is a cut tobacco dryer, the natural parameter values include one or more combinations of tobacco moisture, ambient temperature, and ambient humidity, the device parameter values include one or more combinations of the hot air damper opening, hot air temperature, cylinder rotation speed, moisture exhaust damper opening, and steam pressure of the rehumidifying device, and the device operation state value includes the cut tobacco moisture content.

[0031] In one embodiment, the first change amount of the target operation data value is determined by the following formula:

[0032] where Y is the target operation data value, and the first change amount of the target operation data value Y is Q w is the w-th group of candidate device parameter values, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

[0033] In one embodiment, the second variation of the target operating data value is determined by the following formula: where Y is the target operating data value, and the second variation of the target operating data value Y is Q w is the w-th group of candidate device parameter values, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the w-th group of candidate device parameter values, and λ is the interference parameter value corresponding to the j-th device parameter value.

[0034] In one embodiment, the stability index of the target operating data value under the w-th group of candidate device parameter values is determined by the following formula:

[0035]

[0036] where Y is the target operating data value, S w is the stability index of the target operating data value under the w-th group of candidate device parameter values, Δp i is the sensitivity of the natural parameter p i Δq j is the sensitivity of the device parameter q j is the sensitivity.

[0037] Another aspect of the present invention provides a system including at least one computing device and at least one storage device storing instructions, wherein, when the instructions are run by the at least one computing device, the at least one computing device is caused to execute the above-mentioned device parameter adjustment method.

[0038] Another aspect of the present invention provides a computer-readable storage medium storing instructions, wherein, when the instructions are run by at least one computing device, the at least one computing device is caused to execute the above-mentioned device parameter adjustment method.

[0039] According to the device parameter adjustment method and system of the exemplary embodiments of the present invention, by using machine learning modeling, the machine can automatically mine the rules and patterns in the data, avoiding the defects of over-relying on business experience and expert rules.

[0040] Some other aspects and / or advantages of the general concept of the present invention will be set forth in part in the following description, and some will be obvious from the description, or can be learned through the implementation of the general concept of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Through the following description with reference to the drawings showing exemplary embodiments, the above and other objects and features of the exemplary embodiments of the present invention will become clearer, wherein:

[0042] Figure 1 A flowchart showing a method for adjusting device parameters according to an exemplary embodiment of the present invention;

[0043] Figure 2 A flowchart showing the adjustment of device parameter values according to an exemplary embodiment of the present invention;

[0044] Figure 3 A block diagram showing a device parameter adjustment system according to an exemplary embodiment of the present invention;

[0045] Figure 4 A diagram showing an implementation scenario of device parameter adjustment according to an exemplary embodiment of the present invention. Detailed implementation manners

[0046] Reference will now be made in detail to the embodiments of the present invention. Examples of the embodiments are shown in the accompanying drawings, in which the same reference numerals always refer to the same components. The embodiments will be described below with reference to the accompanying drawings to explain the present invention.

[0047] Figure 1 A flowchart showing a method for adjusting device parameters according to an exemplary embodiment of the present invention.

[0048] Refer to Figure 1 , in step S10, obtain a training data set of historical operation information about the device.

[0049] Each piece of training data includes a natural parameter value, a device parameter value, and a device operation state value at a corresponding moment. Here, the natural parameter value may refer to an environmental parameter value in the environment where the device is located that affects the operation of the device and is collected by a sensor of the device; the device parameter value may refer to a parameter representing the state of the device itself during operation; the device operation state value may refer to a quality evaluation value of the output of the device operation. For example, the failure rate of a production line, the quality grade of the output, product qualification data, etc.

[0050] It can be understood that the environmental parameter value is generally uncontrollable and can be monitored by a sensor. The device parameter value is generally controllable. For example, the device can adjust its corresponding parameter value after receiving an instruction. The quality evaluation value can be a discrete variable or a continuous variable. Taking product qualification data as an example, the product qualification data can be represented by a discrete variable such as "qualified" or "unqualified", or can be represented by a continuous variable such as "qualified rate".

[0051] Taking a tobacco dryer as an example, the natural parameter values of the tobacco dryer may include, but are not limited to, one or more combinations of tobacco moisture, ambient temperature, and ambient humidity. The equipment parameter values of the tobacco dryer may include, but are not limited to, one or more combinations of the opening degree of the hot air damper, hot air temperature, cylinder rotation speed, the opening degree of the moisture exhaust damper, and the steam pressure of the rehumidification equipment. The equipment operating state values of the tobacco dryer may include, but are not limited to, the moisture content of the tobacco and the failure rate of the tobacco dryer.

[0052] After obtaining the initial training data on the historical operation information of the equipment, the obtained initial training data can be cleaned and organized into structured data as training data, and the training data can be stored in the database in multiple copies according to daily partitions (slice tables) or full-scale tables (zipper tables) to form a training dataset.

[0053] It should be noted here that the above equipment can be a physical entity device (such as a tobacco dryer) or a virtual device (such as software that can run on a computer, etc.).

[0054] In step S20, based on the training dataset, a cold start model for predicting the equipment operating state based on natural parameter values and equipment parameter values is trained.

[0055] As an example, the cold start model for predicting the equipment operating state based on natural parameter values and equipment parameter values can be trained by existing training model methods. For example, the constructed cold start model is iteratively updated based on the obtained training data to train the cold start model.

[0056] As an example, assume that the set of natural parameter values of equipment M is P = {p1, p2,..., p m}, and the set of equipment parameter values of equipment M is Q = {q1, q2,..., q n}. Record the actual equipment operating state value Y = M(P, Q, t) during the operation of equipment M, where t represents time. In fact, even under the same natural parameter values and equipment parameter values, Y itself has volatility. Therefore, multiple results within a continuous time period can be collected: Y = {y1, y2,..., y t}. All the training data can be recorded as {(P (i) , Q (i) , Y i )} i＝1，2，...，K , where K is the number of results.

[0057] It can be understood that the cold start model can achieve the following function: for any given set P of natural parameter values and set Q of equipment parameter values of equipment M corresponding to time t0, predict the equipment parameter value Y corresponding to time t0.

[0058] For the training data {(P (i) , Q (i) , Y i )} i＝1，2，...，K , take (P (i) , Q (i) ) as the input variables, and take (Y i ) as the target variable. Optionally, a supervised machine learning algorithm can be used for training when initially training the cold start model.

[0059] In step S30, obtain the current natural parameter values and target operating state values of the device, and determine the candidate device parameter values of the device based on the cold start model, the current natural parameter values, and the target operating state values.

[0060] It can be understood that one of the purposes of the present invention is: for any given set P' of new natural parameter values of device M and the desired device operating state value find the most appropriate set Q of device parameter values such that when device M operates based on the set Q of parameter values, the actual device operating state value Y' of device M is as close as possible to the above-mentioned desired device operating state value That is where Y' i = M(P', Q, t i )x. Therefore, based on the cold start model trained in step 20, it is also necessary to transform it into the solution of an inverse problem.

[0061] Therefore, for the step of determining the candidate device parameter values of the device based on the cold start model, the current natural parameter values, and the target operating state values, in one example, the number of device parameter values of the device can be determined. If the determined number is less than a predetermined threshold, the Cartesian product traversal method can be used to solve the above inverse problem. For example, the following processing can be performed:

[0062] First, combine all possible values of the device parameter values one by one to obtain multiple sets of device parameter values.

[0063] Then, input each set of device parameter values in the multiple sets of device parameter values and the current natural parameter values into the trained cold start model respectively to obtain multiple sets of predicted operation data values.

[0064] Next, compare each set of predicted operation data values in the multiple sets of predicted operation data values with the target operation data values respectively, and select the multiple sets of device parameter values corresponding to the most matching predicted operation data values as the candidate device parameter values.

[0065] In addition, when the quantity is not less than a predetermined threshold, since the computational complexity increases as the quantity increases, there may be a problem of excessive computational load. Therefore, if the quantity is not less than the predetermined threshold, a heuristic search algorithm can be used to calculate and determine the candidate device parameter values of the device. Here, the heuristic search algorithm can include, but is not limited to, simulated annealing, genetic algorithm, particle swarm algorithm, etc.

[0066] Taking the genetic algorithm as an example, the process of calculating and determining the candidate device parameter values of the device will be described below. This process can include the following steps:

[0067] Step (1): Construct the logic of encoding.

[0068] Suppose there are three device parameters: x1, x2, x3. Among them, x1 is a discrete value, for example, the value of x1 is {1, 2, 3}; x2 is a continuous value, for example, the value of x2 is {x2|0 < x2 < 100}; x3 is an enumerated value, for example, the value of x3 is {a, b, c, d}.

[0069] Encode the above three device parameters into binary numerical values. Among them, the discrete value x1 is encoded according to the required minimum number of bits (bit), the continuous value x2 is encoded with equal width of 10 bits, and the enumerated value x3 is encoded with one-hot.

[0070] Step (2): Randomly initialize N seeds to generate corresponding encodings and obtain a generation of population. Where N is a positive integer.

[0071] Step (3): Evaluate and record the matching degree with the target parameter value when the latest obtained generation of population is used as the device parameter value.

[0072] Optionally, use the device parameter value with the matching degree reaching the preset matching degree as the candidate device parameter value.

[0073] Step (4): Copy, mutate, and crossover the N seeds to evolve into the next generation of population.

[0074] Step (5): Iteratively perform Step (3) and Step (4) until the preset end condition is met.

[0075] Among them, meeting the preset end condition can be that the number of iterations reaches the preset number of times, or the number of determined candidate device parameter values reaches the preset value.

[0076] In step S40, select the final device parameter value from the candidate device parameter values of the device, and adjust the device parameters of the device according to the final device parameter value.

[0077] As an example, when the candidate device parameter values include multiple groups of candidate device parameter values, one group of the candidate device parameter values is selected as the device parameter value.

[0078] Here, regarding the step of selecting one group of the candidate device parameter values as the device parameter value, in one example, the following processing can be performed for each group of candidate device parameter values:

[0079] First, determine the first change amount of the target operation data value when adding a corresponding interference parameter value to each current natural parameter value.

[0080] Here, the first change amount of the target operation data value can be determined by the following formula (1):

[0081]

[0082] Where Y is the target operation data value, and the first change amount of the target operation data value Y is Q w is the w-th group of candidate device parameter values, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

[0083] Then, determine the second change amount of the target operation data value when adding a corresponding interference parameter value to each candidate parameter value in this group of candidate device parameter values.

[0084] The second change amount of the target operation data value can be determined by the following formula (2):

[0085]

[0086] Where Y is the target operation data value, and the second change amount of the target operation data value Y is Q w is the w-th group of candidate device parameter values, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the w-th group of candidate device parameter values, and λ is the interference parameter value corresponding to the j-th device parameter value.

[0087] Next, based on the determined first change amount and second change amount of the target operation data value, determine the stability index of the target operation data value under this group of candidate device parameter values.

[0088] Before determining the stability index, the natural parameters of the device and the sensitivity of the device parameters can be obtained. Among them, the set of natural parameters P = {p1, p2,..., p mThe sensitivity of} is ΔP = {Δp1, Δp2,..., Δp m}, where Δp1 is the sensitivity of the natural parameter p1, Δp2 is the sensitivity of the natural parameter p2, and so on.

[0089] The sensitivity of the set of device parameters Q = {q1, q2,..., q n} is ΔQ = {Δq1, Δq2,..., Δq n}, where Δq1 is the sensitivity of the device parameter q1, Δq2 is the sensitivity of the device parameter q2, and so on.

[0090] The stability index of the target operating data value under the w-th group of candidate device parameter values can be determined by the following formula (3):

[0091]

[0092] where Y is the target operating data value, S w is the stability index of the target operating data value under the w-th group of candidate device parameter values, Δp i is the sensitivity of the natural parameter p i and Δq j is the sensitivity of the device parameter q j .

[0093] After obtaining all groups of candidate device parameter values, the candidate device parameter value corresponding to the minimum value among the stability indices of the target operating data values obtained under the multiple groups of candidate device parameter values is used as the device parameter value.

[0094] Here, regarding the step of selecting one of the groups of candidate device parameter values as the device parameter value, in another example, first, interpolation processing can be performed on the multiple groups of candidate device parameter values to obtain the expanded multiple groups of candidate device parameter values. Then, the stability indices of the target operating data under the possible values of the expanded multiple groups of candidate device parameter values are determined. Finally, the candidate device parameter value corresponding to the minimum value among the obtained stability indices is used as the device parameter value.

[0095] In addition, since the operation of the device is related not only to the observable parameters but also to some unobservable parameters, the operating data parameter value of the device operating under the finally selected device parameter value may not exactly match the target operating state value. Therefore, as an example, the method may further additionally include the step of adjusting the device parameter value ( Figure 1 not shown in ).

[0096] Specifically, first, after selecting the final device parameter value, the device can be controlled to operate according to the determined device parameter value.

[0097] Then, the true value of the operation data generated after the operation of the device is collected in real time.

[0098] Next, the true value of the operation data is compared with the target operation data value.

[0099] Finally, based on the comparison result, the device parameter value is adjusted.

[0100] Figure 2 The flowchart showing the adjustment of the device parameter value according to an exemplary embodiment of the present invention is shown.

[0101] Refer to Figure 2 , in step S410, a fine-tuning model is constructed for predicting the comparison result between the true value of the operation data and the target operation data value based on the natural parameter value, the device parameter value, and the device parameter adjustment value.

[0102] In step S420, a fine-tuning training data set is obtained, where the fine-tuning training data set may include the historical natural parameter values, historical device parameter values, historical device parameter adjustment values of the device, and the comparison results between the historical true values of the operation data and the target operation state values.

[0103] In step S430, based on the fine-tuning training data set, the fine-tuning model is iteratively updated.

[0104] In step S440, based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the true value of the operation data and the target operation state value, the device parameter adjustment value is determined.

[0105] In step S450, based on the determined device parameter value and the device parameter adjustment value, the adjusted device parameter value is determined.

[0106] Here, the method for determining the adjusted device parameter value is similar to the above steps for selecting the final device parameter value, and the difference between the two lies in the training data of the model and the target variable.

[0107] The training data of the fine-tuning model can be recorded as {(P (i) , Q (i) , ΔQ (i) ); ΔY i} i＝1，2，...，K . ΔQ (i) is the set of device parameter adjustment values, and ΔY i is the set of comparison results between the true value of the operation data and the target operation data value.

[0108] First, ensure that the fine-tuning model can achieve the following functions: for any given set P of natural parameter values and set Q of device parameter values of device M corresponding to time t0, a device parameter adjustment value ΔQ is made to the set Q of device parameter values(i) Adjustment can be made to the device parameter value ΔY corresponding to time t0. Therefore, when training the fine-tuning model (P (i) , Q (i) , ΔQ (i) ) are used as input variables, and ΔY i is used as the target variable.

[0109] It can be understood that one of the objectives of the present invention is: for any given set p′ of new natural parameter values of device M, set Q′ of parameter values, and comparison result ΔY between the true value of the operating data and the target operating data value, to find the most appropriate set ΔQ of device parameter adjustment values (i) , such that after the device M makes an adjustment of a device parameter adjustment value ΔQ to the set Q of the parameter values (i) , the actual device operating state value Y′ is as close as possible to the desired device operating state value Therefore, the fine-tuning model obtained by training based on step S430 also needs to be transformed into the solution of an inverse problem.

[0110] It can be understood that the principles of steps S410 to S450 are the same as those of steps S10 to S40 above. The descriptions of steps S410 to S450 are not repeated here.

[0111] According to the device parameter adjustment method of the exemplary embodiment of the present invention, by using machine learning modeling to enable the machine to automatically discover the laws and patterns in the data, the defects of over-reliance on business experience and expert rules are avoided.

[0112] Figure 3 is a block diagram of the device parameter adjustment system of the exemplary embodiment of the present invention.

[0113] Brief descriptions will be given below of the functional units that the device parameter adjustment system may have and the operations that each functional unit can perform. For the detailed parts involved, reference can be made to the relevant descriptions above, which will not be repeated here.

[0114] Refer to Figure 3, the device parameter adjustment system of the exemplary embodiment of the present invention includes a training data acquisition module 510, a model training module 520, a candidate parameter determination module 530, and a device parameter adjustment module 540; the training data acquisition module 510 is configured to acquire a training data set regarding the historical operation information of the device, wherein each piece of training data includes a natural parameter value, a device parameter value, and a device operation state value at a corresponding moment; the model training module 520 is configured to train a cold start model for predicting the device operation state based on the natural parameter value and the device parameter value based on the training data set; the candidate parameter determination module 530 is configured to acquire the current natural parameter value and the target operation state value regarding the device, and determine the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operation state value; the device parameter adjustment module 540 is configured to select the final device parameter value from the candidate device parameter values of the device, and adjust the device parameter of the device according to the final device parameter value.

[0115] In one implementation, the candidate parameter determination module 530 is configured to: determine the number of device parameter values regarding the device; if the number is less than a predetermined threshold, perform the following processing: combine all possible values of the device parameter values one by one to obtain multiple groups of device parameter values, respectively input each group of device parameter values in the multiple groups of device parameter values and the current natural parameter value into the trained cold start model to obtain multiple groups of predicted operation data values, respectively compare each group of predicted operation data values in the multiple groups of predicted operation data values with the target operation data value, and select the multiple groups of device parameter values corresponding to the most matching predicted operation data value as the candidate device parameter values; if the number is not less than the predetermined threshold, use a heuristic search algorithm to calculate and determine the candidate device parameter value of the device.

[0116] In one implementation, the device parameter adjustment module 540 is configured to: when the candidate device parameter values include multiple groups of candidate device parameter values, select one group of candidate device parameter values as the device parameter value.

[0117] In one implementation, the device parameter adjustment module 540 is configured to perform the following processing for each group of candidate device parameter values: determine the first change amount of the target operation data value when adding a corresponding interference parameter value to each current natural parameter value; determine the second change amount of the target operation data value when adding a corresponding interference parameter value to each candidate parameter value in the group of candidate device parameter values; based on the determined first change amount and second change amount of the target operation data value, determine the stability index of the target operation data value under the group of candidate device parameter values, and use the candidate device parameter value corresponding to the minimum value among the stability indexes of the target operation data values obtained under the multiple groups of candidate device parameter values as the device parameter value.

[0118] In one embodiment, the device parameter adjustment module 540 is configured to: perform interpolation processing on the multiple sets of candidate device parameter values to obtain the expanded multiple sets of candidate device parameter values; determine the stability index of the target operation data under the possible values of the expanded multiple sets of candidate device parameter values; and use the candidate device parameter value corresponding to the minimum value in the obtained stability indexes as the device parameter value.

[0119] In one embodiment, the system further includes a device operation control module (not shown in the figure), and the device operation control module is configured to: control the device to operate according to the determined device parameter value; collect the real value of the operation data generated after the device operates in real time; compare the real value of the operation data with the target operation data value; and adjust the device parameter value based on the comparison result.

[0120] In one embodiment, the device operation control module is configured to: construct a fine-tuning model for predicting the comparison result between the real value of the operation data and the target operation data value based on the natural parameter value, the device parameter value, and the device parameter adjustment value; obtain a fine-tuning training data set, where the fine-tuning training data set includes the historical natural parameter values, historical device parameter values, historical device parameter adjustment values of the device, and the comparison results between the historical real values of the operation data and the target operation state values; iteratively update the fine-tuning model based on the fine-tuning training data set; determine the device parameter adjustment value based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the real value of the operation data and the target operation data value; and determine the adjusted device parameter value based on the determined device parameter value and the device parameter adjustment value.

[0121] In one embodiment, the device is a cut tobacco dryer, the natural parameter values include one or more combinations of tobacco moisture, ambient temperature, and ambient humidity, the device parameter values include one or more combinations of the opening degree of the hot air damper, the hot air temperature, the cylinder rotation speed, the opening degree of the moisture exhaust damper, and the steam pressure of the rehumidifying device, and the device operation state value includes the moisture content of the cut tobacco.

[0122] In one embodiment, the first change amount of the target operation data value is determined by the following formula:

[0123] where Y is the target operation data value, and the first change amount of the target operation data value Y is Q w is the w-th set of candidate device parameter values, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

[0124] In one embodiment, the second variation of the target operating data value is determined by the following formula: Where Y is the target operating data value, and the second variation of the target operating data value Y is Q w is the w-th group of candidate device parameter values, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the w-th group of candidate device parameter values, and λ is the interference parameter value corresponding to the j-th device parameter value.

[0125] In one embodiment, the stability index of the target operating data value under the w-th group of candidate device parameter values is determined by the following formula:

[0126]

[0127] Where Y is the target operating data value, S w is the stability index of the target operating data value under the w-th group of candidate device parameter values, Δp i is the sensitivity of the natural parameter p i Δq j is the sensitivity of the device parameter q j is the sensitivity.

[0128] According to the device parameter adjustment system of the exemplary embodiment of the present invention, by using machine learning modeling, the machine can automatically mine the rules and patterns in the data, avoiding the defect of over-relying on business experience and expert rules.

[0129] Figure 4 is a diagram showing an implementation scenario of device parameter adjustment according to the exemplary embodiment of the present invention. In this implementation scenario, at least one electronic device 610 and at least one controlled device 620 may be included. The at least one electronic device 610 may include various types of electronic devices such as computer devices, smart phones, tablet computers, laptop computers, servers, etc., which can be used to execute the device parameter adjustment method.

[0130] The electronic device 610 can be communicatively connected to the controlled device 620. The electronic device 610 obtains a training data set of historical operation information of the controlled device 620. Each piece of training data includes natural parameter values, device parameter values, and device operation status values at corresponding times. The electronic device 610 trains a cold start model for predicting the device operation status based on the natural parameter values and device parameter values, based on the training data set. The electronic device 610 obtains the current natural parameter values and target operation status values of the controlled device 620, and determines candidate device parameter values of the controlled device 620 based on the cold start model, the current natural parameter values, and the target operation status values. The electronic device 610 selects the final device parameter values from the candidate device parameter values of the controlled device 620, and adjusts the device parameters of the controlled device 620 according to the final device parameter values.

[0131] The electronic device 610 controls the controlled device 620 to operate according to the determined device parameter values; the electronic device 610 collects the true values of the operation data generated after the controlled device 620 operates in real time; the electronic device 610 compares the true values of the operation data with the target operation data values; the electronic device 610 adjusts the parameter values of the controlled device 620 based on the comparison result.

[0132] The training data set of the historical operation information can be input by the user to the electronic device 610, and the training data set of the historical operation information is also sent to the electronic device 610 by other devices.

[0133] The electronic device 610 can obtain the current natural parameter values of the device through the sensors on the controlled device 620. The target operation status value can be a default value, or can be set by the user based on the electronic device 610, or can be sent to the electronic device 610 by other devices. The electronic device 610 directly outputs the determined final device parameter values to the controlled device 620.

[0134] According to the device parameter adjustment method and system of the exemplary embodiment of the present invention, by using machine learning modeling, the machine automatically mines the rules and patterns in the data, avoiding the defects of over-reliance on business experience and expert rules.

[0135] The above has been referred to Figures 1 to 3 described the device parameter adjustment method and system according to the exemplary embodiment of the present invention.

[0136] Figure 3Each unit in the shown device parameter adjustment system can be configured as software, hardware, firmware, or any combination of the above for performing specific functions. For example, each unit can correspond to an application-specific integrated circuit, or to pure software code, or to a module combining software and hardware. In addition, one or more functions implemented by each unit can also be uniformly executed by components in a physical entity device (such as a processor, a client, or a server, etc.).

[0137] In addition, referring to Figure 1 The described device parameter adjustment method can be implemented by a program (or instruction) recorded on a computer-readable storage medium. For example, according to an exemplary embodiment of the present invention, a computer-readable storage medium storing instructions can be provided, wherein when the instructions are run by at least one computing device, the at least one computing device is caused to execute the device parameter adjustment method according to an exemplary embodiment of the present invention.

[0138] The computer program in the above computer-readable storage medium can run in an environment deployed in computer devices such as clients, hosts, proxy devices, servers, etc. It should be noted that the computer program can also be used to execute additional steps other than the above steps or perform more specific processing when executing the above steps. The content of these additional steps and further processing has been mentioned during the description of the related method with reference to Figure 1 Therefore, in order to avoid repetition, it will not be elaborated here.

[0139] It should be noted that each unit in the device parameter adjustment system according to an exemplary embodiment of the present invention can completely rely on the running of the computer program to implement the corresponding functions, that is, each unit corresponds to each step in the functional architecture of the computer program, so that the entire system is called through a dedicated software package (such as a lib library) to implement the corresponding functions.

[0140] On the other hand, Figure 3 Each of the shown units can also be implemented by hardware, software, firmware, middleware, microcode, or any combination thereof. When implemented by software, firmware, middleware, or microcode, the program code or code segment for performing the corresponding operations can be stored in a computer-readable medium such as a storage medium, so that the processor can execute the corresponding operations by reading and running the corresponding program code or code segment.

[0141] For example, an exemplary embodiment of the present invention can also be implemented as a computing device, which includes a storage component and a processor. The storage component stores a set of computer-executable instructions. When the set of computer-executable instructions is executed by the processor, the device parameter adjustment method according to an exemplary embodiment of the present invention is executed.

[0142] Specifically, the computing device can be deployed in a server or a client, or on a node device in a distributed network environment. In addition, the computing device can be a PC computer, a tablet device, a personal digital assistant, a smart phone, a web application, or other devices capable of executing the above instruction set.

[0143] Here, the computing device does not have to be a single computing device, but can also be any collection of devices or circuits capable of executing the above instructions (or instruction sets) individually or jointly. The computing device can also be part of an integrated control system or a system manager, or can be configured as a portable electronic device that interfaces with a local or remote (e.g., via wireless transmission).

[0144] In the computing device, the processor can include a central processing unit (CPU), a graphics processing unit (GPU), a programmable logic device, a dedicated processor system, a microcontroller, or a microprocessor. By way of example and not limitation, the processor can also include an analog processor, a digital processor, a microprocessor, a multi-core processor, a processor array, a network processor, etc.

[0145] Certain operations described in the device parameter adjustment method according to an exemplary embodiment of the present invention can be implemented in software, certain operations can be implemented in hardware, and in addition, these operations can also be implemented in a combination of software and hardware.

[0146] The processor can run instructions or code stored in one of the storage components, where the storage component can also store data. The instructions and data can also be sent and received via the network interface device over the network, where the network interface device can employ any known transmission protocol.

[0147] The storage component can be integrated with the processor, for example, by arranging RAM or flash memory within an integrated circuit microprocessor, etc. In addition, the storage component can include independent devices, such as external disk drives, storage arrays, or other storage devices that can be used by any database system. The storage component and the processor can be operatively coupled, or can communicate with each other, for example, via I / O ports, network connections, etc., such that the processor can read files stored in the storage component.

[0148] In addition, the computing device can also include a video display (such as a liquid crystal display) and a user interaction interface (such as a keyboard, a mouse, a touch input device, etc.). All components of the computing device can be connected to each other via a bus and / or a network.

[0149] The device parameter adjustment method according to an exemplary embodiment of the present invention can be described as various interconnected or coupled functional blocks or functional diagrams. However, these functional blocks or functional diagrams can be equally integrated into a single logical device or operate with non-exact boundaries.

[0150] Therefore, referring to Figure 1 the device parameter adjustment method described can be implemented by a system including at least one computing device and at least one storage device storing instructions.

[0151] According to an exemplary embodiment of the present invention, at least one computing device is a computing device for executing the device parameter adjustment method according to an exemplary embodiment of the present invention, and a set of computer-executable instructions is stored in the storage device. When the set of computer-executable instructions is executed by at least one computing device, the device parameter adjustment method described in Figure 1 is executed.

[0152] The above describes various exemplary embodiments of the present invention. It should be understood that the above description is merely exemplary and not exhaustive, and the present invention is not limited to the disclosed exemplary embodiments. Many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the present invention. Therefore, the protection scope of the present invention should be defined by the scope of the claims.

[0153] The above describes various exemplary embodiments of the present invention. It should be understood that the above description is merely exemplary and not exhaustive, and the present invention is not limited to the disclosed exemplary embodiments. Many modifications and variations are obvious to those of ordinary skill in the art without departing from the scope and spirit of the present invention. Therefore, the protection scope of the present invention should be defined by the scope of the claims.

Claims

1. A method for adjusting device parameters, wherein, The method includes: Obtaining a training data set of historical operation information of the device, where each piece of training data includes natural parameter values, device parameter values, and device operation state values at corresponding moments, and the natural parameter values are environmental parameter values in the environment where the device is located that affect the operation of the device; Training a cold start model for predicting the device operation state based on the natural parameter values and device parameter values based on the training data set; Obtaining the current natural parameter values and target operation state values of the device, and determining the candidate device parameter values of the device based on the cold start model, the current natural parameter values, and the target operation state values; Selecting the final device parameter values from the candidate device parameter values of the device, and adjusting the device parameters of the device according to the final device parameter values; Among them, the step of selecting the final device parameter values from the candidate device parameter values of the device includes: When the candidate device parameter values include multiple sets of candidate device parameter values, the following processing is performed for each set of candidate device parameter values: determining the first change amount of the target operation state value when adding a corresponding interference parameter value to each current natural parameter value; determining the second change amount of the target operation state value when adding a corresponding interference parameter value to each candidate parameter value in this set of candidate device parameter values; based on the determined first change amount and second change amount of the target operation state value, determining the stability index of the target operation state value under this set of candidate device parameter values; Taking the candidate device parameter value corresponding to the minimum value among the stability indexes of the target operation state values obtained under the multiple sets of candidate device parameter values as the device parameter value.

2. The method according to claim 1, wherein The step of determining the candidate device parameter values of the device based on the cold start model, the current natural parameter values, and the target operation state values includes: Determining the number of device parameter values of the device; If the number is less than a predetermined threshold, the following processing is performed: Combining all possible values of the device parameter values one by one to obtain multiple sets of device parameter values, Inputting each set of device parameter values in the multiple sets of device parameter values and the current natural parameter values into the trained cold start model respectively to obtain multiple sets of predicted operation state values, Comparing each set of predicted operation state values in the multiple sets of predicted operation state values with the target operation state value respectively, Selecting the multiple sets of device parameter values corresponding to the most matching predicted operation state value as the candidate device parameter values; If the number is not less than the predetermined threshold, using a heuristic search algorithm to calculate and determine the candidate device parameter values of the device.

3. The method according to claim 1, wherein The step of selecting one set of candidate device parameter values as the device parameter value includes: Performing interpolation processing on the multiple sets of candidate device parameter values to obtain the expanded multiple sets of candidate device parameter values; Determining the stability index of the target operation state under the possible values of the expanded multiple sets of candidate device parameter values; Taking the candidate device parameter value corresponding to the minimum value among the obtained stability indexes as the device parameter value.

4. The method according to claim 1, wherein The method further includes: Controlling the device to operate according to the determined device parameter values; Real-time collecting the true values of the operation states generated after the device operates; Compare the true value of the operating state with the target operating state value; Based on the comparison result, adjust the device parameter value.

5. The method according to claim 4, wherein, The steps of adjusting the device parameter value include: Construct a fine-tuning model for predicting the comparison result between the true value of the operating state and the target operating state value based on the natural parameter value, the device parameter value, and the device parameter adjustment value; Obtain a fine-tuning training data set, where the fine-tuning training data set includes the historical natural parameter value, the historical device parameter value, the historical device parameter adjustment value, and the comparison result between the true value of the historical operating state and the target operating state value of the device; Based on the fine-tuning training data set, iteratively update the fine-tuning model; Based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the true value of the operating state and the target operating state value, determine the device parameter adjustment value; Based on the determined device parameter value and the device parameter adjustment value, determine the adjusted device parameter value.

6. The method according to claim 1, wherein, The device is a cut tobacco dryer, The natural parameter value includes a combination of one or more of the tobacco moisture content, the ambient temperature, and the ambient humidity, The device parameter value includes a combination of one or more of the hot air damper opening, the hot air temperature, the cylinder rotation speed, the moisture exhaust damper opening, and the steam pressure of the rehumidifying device, The device operating state value includes the cut tobacco moisture content.

7. The method according to claim 1, wherein Determine the first change amount of the target operating state value through the following formula: Among them, Y is the target operating state value, and the first change amount of the target operating state value Y is Q w is the candidate device parameter value of the w-th group, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

8. The method according to claim 7, wherein, Determine the second change amount of the target operating state value through the following formula: where Y is the target operating state value, and the second change amount of the target operating state value Y is Q w is the candidate device parameter value of the w-th group, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the candidate device parameter values of the w-th group, and λ is the interference parameter value corresponding to the j-th device parameter value.

9. The method according to claim 8, wherein Determine the stability index of the target operating state value under the w-th group of candidate device parameter values through the following formula: Among them, Y is the target operating state value, and S w is the stability index of the target operating state value under the w-th group of candidate device parameter values, and Δp i is the sensitivity of the natural parameter p i and Δq j is the sensitivity of the device parameter q j .

10. A device parameter adjustment system, wherein, The system includes: A training data acquisition module configured to acquire a training data set of historical operating information of the device, where each piece of training data includes the natural parameter value, the device parameter value, and the device operating state value at the corresponding moment, and the natural parameter value is the environmental parameter value in the environment where the device is located that affects the operation of the device; A model training module configured to train a cold start model for predicting the device operating state based on the natural parameter value and the device parameter value based on the training data set; A candidate parameter determination module configured to acquire the current natural parameter value and the target operating state value of the device, and determine the candidate device parameter value of the device based on the cold start model, the current natural parameter value, and the target operating state value; A device parameter adjustment module configured to select the final device parameter value from the candidate device parameter values of the device and adjust the device parameter of the device according to the final device parameter value; Wherein, the device parameter adjustment module is configured to: When the candidate device parameter values include multiple sets of candidate device parameter values, the following processing is performed for each set of candidate device parameter values: Determine the first change amount of the target operating state value when adding a corresponding interference parameter value to each current natural parameter value; Determine the second change amount of the target operating state value when adding a corresponding interference parameter value to each candidate parameter value in this set of candidate device parameter values; Based on the determined first change amount and second change amount of the target operating state value, determine the stability index of the target operating state value under this set of candidate device parameter values; Use the candidate device parameter value corresponding to the minimum value among the stability indexes of the target operating state values obtained under the multiple sets of candidate device parameter values as the device parameter value.

11. The system according to claim 10, wherein, The candidate parameter determination module is configured to: Determine the number of device parameter values regarding the device; If the number is less than a predetermined threshold, perform the following processing: Combine all possible values of the device parameter values one by one to obtain multiple sets of device parameter values, Input each set of device parameter values in the multiple sets of device parameter values and the current natural parameter value into the trained cold start model respectively to obtain multiple sets of predicted operating state values, Compare each set of predicted operating state values in the multiple sets of predicted operating state values with the target operating state value respectively, Select the multiple sets of device parameter values corresponding to the most matching predicted operating state value as the candidate device parameter values; If the number is not less than the predetermined threshold, use a heuristic search algorithm to calculate and determine the candidate device parameter values of the device.

12. The system according to claim 10, wherein, The device parameter adjustment module is configured to: Perform interpolation processing on the multiple sets of candidate device parameter values to obtain the expanded multiple sets of candidate device parameter values; Determine the stability index of the target operating state under the possible values of the expanded multiple sets of candidate device parameter values; Use the candidate device parameter value corresponding to the minimum value among the obtained stability indexes as the device parameter value.

13. The system according to claim 10, wherein, The system further includes a device operation control module, and the device operation control module is configured to: Control the device to operate according to the determined device parameter value; Collect the real value of the operating state generated after the device operates in real time; Compare the real value of the operating state with the target operating state value; Adjust the device parameter value based on the comparison result.

14. The system according to claim 13, wherein, The device operation control module is configured to: Construct a fine-tuning model for predicting the comparison result between the real value of the operating state and the target operating state value based on the natural parameter value, the device parameter value, and the device parameter adjustment value; Obtain a fine-tuning training data set, where the fine-tuning training data set includes the historical natural parameter value, the historical device parameter value, the historical device parameter adjustment value, and the comparison result between the historical real value of the operating state and the target operating state value regarding the device; Iteratively update the fine-tuning model based on the fine-tuning training data set; Determine the device parameter adjustment value based on the iteratively updated fine-tuning model, the current natural parameter value, the device parameter value, and the comparison result between the real value of the operating state and the target operating state value; Determine the adjusted device parameter value based on the determined device parameter value and the device parameter adjustment value.

15. The system according to claim 10, wherein, The device is a cut tobacco dryer, The natural parameter values include a combination of one or more of the tobacco moisture, ambient temperature, and ambient humidity, The device parameter values include a combination of one or more of the hot air damper opening degree, hot air temperature, cylinder rotation speed, moisture exhaust damper opening degree, and steam pressure of the rehumidifying device, The device operating state value includes the moisture content of the cut tobacco.

16. The system according to claim 10, wherein The first change amount of the target operating state value is determined by the following formula: Among them, Y is the target operating state value, and the first change amount of the target operating state value Y is Q w is the candidate device parameter value of the w-th group, is the objective function indicated by the cold start model, p i is the i-th current natural parameter value, and ε is the interference parameter value corresponding to the i-th current natural parameter value.

17. The system according to claim 16, wherein, The second change amount of the target operating state value is determined by the following formula: Among them, Y is the target operating state value, and the second change amount of the target operating state value Y is Q w is the candidate device parameter value of the w-th group, is the objective function indicated by the cold start model, q j is the j-th device parameter value in the candidate device parameter values of the w-th group, and λ is the interference parameter value corresponding to the j-th device parameter value.

18. The system according to claim 17, wherein The stability index of the target operating state value under the w-th group of candidate device parameter values is determined by the following formula: Among them, Y is the target operating state value, and S w is the stability index of the target operating state value under the candidate device parameter values of the w-th group, and Δp i is the sensitivity of the natural parameter p i and Δq j is the sensitivity of the device parameter q j .

19. A system comprising at least one computing device and at least one storage device storing instructions, wherein, When the instruction is run by the at least one computing device, it causes the at least one computing device to execute the device parameter adjustment method according to any one of claims 1 to 9.

20. A computer-readable storage medium for storing instructions, wherein, When the instruction is run by at least one computing device, it causes the at least one computing device to execute the device parameter adjustment method according to any one of claims 1 to 9.

Images