Method and system for detecting quality of cooling liquid of compressor with high cleanliness and long service life

Abstract

The invention provides a high-cleanliness long-life compressor coolant quality detection method and system, and relates to the technical field of performance detection.The method comprises the steps that a clearing agent is added into base oil to obtain first additive oil, and an oil sludge control score, an anti-corrosion score and a sediment score are tested; before and after the antioxidant is added, a first oxidation induction period score and a first thermal weight loss performance score are obtained, a second oxidation induction period score and a second thermal weight loss performance score are obtained, and then third additive oil is screened; after the anti-wear agent is added, the anti-wear score and the industrial performance index are tested; and training the cooling liquid performance prediction model so as to carry out simulation test on more proportions. According to the invention, in the testing process, the cooling liquid performance prediction model is trained by following the preparation process of the clearing agent, the antioxidant and the anti-wear agent, so that the performance of the cooling liquid with more proportions can be predicted through the cooling liquid performance prediction model, the testing difficulty and cost can be reduced, and the optimal proportion can be found.

Description

Technical Field

[0001] The present invention relates to the technical field of performance detection, and in particular to a method and system for detecting the quality of a high-cleanliness and long-life compressor coolant. Background Art

[0002] Coolants used in mining air compressors are prone to coking over long periods of use. For example, with mineral oil-based coolants, some components may chemically react over time and with temperature fluctuations, forming coking substances such as carbon deposits and colloids. These deposits adhere to the compressor's internal components, affecting performance such as heat dissipation and lubrication, thereby shortening the equipment's lifespan. With the continuous development of industrial production, air compressors are increasingly used, operating for long periods of time and under heavy loads. The demand for coolants that ensure stable, long-term operation and reduce maintenance costs is increasing, and existing coolants prone to coking are no longer able to meet this demand.

[0003] However, when exploring coolant formulas that can meet the needs, the performance of multiple chemical substances is difficult to coordinate well. Therefore, when the performance requirements are high and a large number of additives are required, it is difficult to find a suitable ratio.

[0004] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the Invention

[0005] The present invention provides a method and system for detecting the quality of a high-cleanliness, long-life compressor coolant, which can solve the technical problem in related technologies of difficulty in finding a suitable ratio when high performance requirements and a large number of required additives are present.

[0006] According to a first aspect of the present invention, a method for detecting the quality of a high-cleanliness, long-life compressor coolant is provided, comprising:

[0007] adding a second preset volume of detergent to multiple barrels of base oil of a first preset volume, respectively, to obtain multiple barrels of first added oil, wherein the detergent includes multiple detergents, and the proportions of the multiple detergents in each barrel of base oil are different;

[0008] Sampling each barrel of first-added oil, and subjecting the first samples obtained from the sampling to sludge control test, corrosion test, and sediment test, respectively, to obtain a sludge control score, corrosion resistance score, and sediment score for each barrel of first-added oil;

[0009] Sampling each barrel of the first added oil, and testing the first oxidation induction period score and the first thermal weight loss performance score of each second sample;

[0010] Each barrel of the first added oil is divided into a plurality of equal portions, a third preset volume of an antioxidant is added to each portion of the first added oil to obtain a plurality of portions of the second added oil, and the second added oil is sampled to obtain third samples, and a second oxidation induction period score and a second thermal weight loss performance score are determined for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different from each other;

[0011] Screening a third added oil from a plurality of second added oils according to the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score;

[0012] dividing each portion of the third added oil into a plurality of portions, and adding a fourth preset volume of an anti-wear agent to each portion of the third added oil to obtain a plurality of portions of the fourth added oil, wherein the anti-wear agent comprises a plurality of anti-wear substances, and the ratios of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil are different from each other;

[0013] Determining an anti-wear score for each fourth added oil and determining an industrial performance index for each fourth added oil;

[0014] Obtaining a sludge control score, an anti-corrosion score, a deposit score, a second oxidation induction period score, a second thermal weight loss performance score, an anti-wear score, and an industrial performance index of each fourth added oil, and training a coolant performance prediction model to obtain a trained coolant performance prediction model;

[0015] Using the trained coolant performance prediction model, simulated tests were conducted on detergents, antioxidants, and anti-wear agents to obtain test results.

[0016] According to the present invention, based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score, a third added oil is screened out from multiple second added oils, including:

[0017] Among the plurality of second added oils, selecting a candidate second added oil having a second oxidation induction period score higher than a preset induction period score lower limit and a second thermal weight loss performance score higher than a preset thermal weight loss performance score lower limit;

[0018] Determining an oxidation induction period score increase based on the first oxidation induction period score and the second oxidation induction period score of the second added oil, and determining a thermal weight loss performance increase based on the first thermal weight loss performance score and the second thermal weight loss performance score of the second added oil;

[0019] The synergistic performance score is obtained by weighted summing the increase in the oxidation induction period score, the increase in the thermal weight loss performance, the sludge control score, the corrosion resistance score, and the deposit score.

[0020] According to the synergistic performance score, the third added oil is selected from multiple candidate second added oils.

[0021] According to the present invention, the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score and industrial performance index of each fourth added oil are obtained, and a coolant performance prediction model is trained to obtain a trained coolant performance prediction model, including:

[0022] Obtaining the amount and ratio of detergent, antioxidant, and antiwear agent in each portion of the fourth added oil;

[0023] Inputting the amount and ratio of detergent added into the first coding level to obtain a detergent addition feature vector;

[0024] Inputting the amount and ratio of the antioxidant into the second coding level to obtain the antioxidant addition feature vector;

[0025] Input the amount and ratio of the anti-wear agent into the third coding level to obtain the anti-wear agent addition feature vector;

[0026] Input the detergent addition feature vector into the first fully connected layer to obtain the predicted sludge control score, the predicted corrosion protection score, and the predicted deposit score;

[0027] Input the detergent addition feature vector and the antioxidant addition feature vector into the first attention mechanism layer to obtain the antioxidant feature vector;

[0028] The antioxidant feature vector is input into the second fully connected layer to obtain the predicted oxidation induction period score and the predicted thermal weight loss performance score;

[0029] Input the antioxidant feature vector and the anti-wear agent addition feature vector into the second attention mechanism layer to obtain the anti-wear feature vector;

[0030] The anti-wear feature vector is input into the third fully connected layer to obtain the predicted anti-wear score;

[0031] Input the detergent addition feature vector, antioxidant addition feature vector, and antiwear agent addition feature vector into the third attention mechanism layer to obtain the industrial performance feature vector;

[0032] Input the industrial performance feature vector into the fourth fully connected layer to obtain the predicted industrial performance index;

[0033] Obtaining a loss function of a coolant performance prediction model based on the predicted sludge control score, the predicted corrosion protection score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score, and the predicted industrial performance index, as well as the sludge control score, the corrosion protection score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score, and the industrial performance index;

[0034] The coolant performance prediction model is trained according to the loss function of the coolant performance prediction model to obtain a trained coolant performance prediction model.

[0035] According to the present invention, based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, a loss function of the coolant performance prediction model is obtained, including:

[0036] Obtaining a detergency loss function based on the predicted sludge control score, the predicted corrosion protection score, the sludge control score, the corrosion protection score, the predicted deposit score, and the deposit score;

[0037] Obtaining an antioxidant performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, and the second thermal weight loss performance score;

[0038] Obtaining an anti-wear performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, the second thermal weight loss performance score, the predicted anti-wear score, and the anti-wear score;

[0039] Obtaining an industrial performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score, and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score, and the industrial performance index;

[0040] The loss function of the coolant performance prediction model is determined based on the cleaning performance loss function, the anti-oxidation performance loss function, the anti-wear performance loss function and the industrial performance loss function.

[0041] According to the present invention, a detergency loss function is obtained based on the predicted sludge control score, the predicted corrosion protection score, the sludge control score, the corrosion protection score, the predicted deposit score, and the deposit score, including:

[0042] According to the formula , Obtaining a clean performance loss function ,in, To predict the sludge control score, Score sludge control, To predict the corrosion resistance score, Score for corrosion resistance, To score the predicted sediment, Score the sediment, 、 and The preset weights.

[0043] According to the present invention, based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score and the second thermal weight loss performance score, an antioxidant performance loss function is obtained, including:

[0044] According to the formula ,

[0045] Obtaining antioxidant performance loss function ,in, To predict the oxidative induction period score, Score the second oxidation induction period, To predict the thermal weight loss performance score, For the second thermal weight loss performance score, and is the preset weight, and max is the maximum value function.

[0046] According to the present invention, based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, the second thermal weight loss performance score, the predicted anti-wear score and the anti-wear score, an anti-wear performance loss function is obtained, including:

[0047] According to the formula , Obtain anti-wear performance loss function ,in, To predict the anti-wear score, Rating for wear resistance.

[0048] According to the present invention, based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, an industrial performance loss function is obtained, including:

[0049] According to the formula , Obtaining industrial performance loss function ,in, is the i-th industrial performance index, is the i-th predicted industrial performance indicator, n is the number of types of industrial performance indicators, i≤n, and both i and n are positive integers.

[0050] According to a second aspect of the present invention, a high-cleanliness and long-life compressor coolant quality detection system is provided, comprising:

[0051] a first oil addition module for adding a second preset volume of detergent to a plurality of barrels of base oil of a first preset volume, respectively, to obtain a plurality of barrels of first added oil, wherein the detergent comprises a plurality of detergent substances, and the proportions of the plurality of detergent substances in the detergent added to each barrel of base oil are different;

[0052] a cleanliness test module, for sampling each barrel of first-added oil, and performing sludge control test, corrosion test, and sediment test on the first sample obtained, to obtain a sludge control score, corrosion resistance score, and sediment score for each barrel of first-added oil;

[0053] a first oxidation test module, for sampling each barrel of the first added oil and detecting a first oxidation induction period score and a first thermal weight loss performance score of each second sample;

[0054] a second oxidation test module, configured to divide each barrel of the first added oil into a plurality of equal portions, add a third preset volume of an antioxidant to each portion of the first added oil to obtain a plurality of portions of the second added oil, sample the second added oil to obtain third samples, and determine a second oxidation induction period score and a second thermal weight loss performance score for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different;

[0055] a screening module for screening a third added oil from a plurality of second added oils based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score;

[0056] a fourth oil addition module, configured to divide each portion of the third added oil into a plurality of portions, and to add a fourth preset volume of anti-wear agent to each portion of the third added oil to obtain a plurality of fourth added oil portions, wherein the anti-wear agent comprises a plurality of anti-wear substances, and the ratios of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil are different from each other;

[0057] an anti-wear test module for determining an anti-wear score for each portion of the fourth additive oil and determining an industrial performance index for each portion of the fourth additive oil;

[0058] A training module is used to obtain a sludge control score, an anti-corrosion score, a deposit score, a second oxidation induction period score, a second thermal weight loss performance score, an anti-wear score, and an industrial performance index of each fourth added oil, and to train a coolant performance prediction model to obtain a trained coolant performance prediction model;

[0059] The simulation test module is used to simulate the test of detergents, antioxidants and anti-wear agents using the trained coolant performance prediction model to obtain test results.

[0060] By adopting the above technical solution, the present invention can achieve the following technical effects:

[0061] According to the present invention, various detergent formulations can be tested for sludge control, corrosion, and deposits, and antioxidants can be tested for their effects on oxidation induction period and thermal weight loss performance. The antiwear properties of antiwear agents can also be tested. During the testing process, the formulation processes of the detergents, antioxidants, and antiwear agents are followed, and a coolant performance prediction model is trained based on these formulation processes. This allows the coolant performance prediction model to predict the performance of a wider range of coolant formulations, thereby reducing testing difficulty and cost and facilitating the search for optimized formulations. When constructing the coolant performance prediction model, the model's processing flow can be aligned with the testing flow for various coolant test indicators, enabling the model to better simulate the coolant testing process and improving the model's structural scientificity and prediction accuracy. When determining the cleaning performance loss function, the measured values ​​of each score can be used as training labels, and the error between the predicted scores output by the model and the training labels is used to form the loss function. This allows the model output value to be closer to the measured value during training, improving model accuracy. When determining the antioxidant performance loss function and the anti-wear performance loss function, the main source of error can be determined by taking the maximum function, thereby judging whether it is necessary to conduct separate intensive training on the previous branches, so as to reduce the errors of each branch while reducing the overall error, thereby improving the pertinence and efficiency of training and improving the training quality. It is also possible to strengthen the training of the branches corresponding to the first coding level, the second coding level, and the third coding level by constructing an industrial performance loss function. It is also possible to reduce the average value of the relative error between the predicted industrial performance index and the industrial performance index during training to train the first coding level, the second coding level, the third coding level, the third attention mechanism level, and the fourth fully connected layer, thereby further strengthening the training of the first coding level, the second coding level, and the third coding level. Through multiple training of the first coding level, the second coding level, and the third coding level, the accuracy of feature information extraction of the first coding level, the second coding level, and the third coding level is comprehensively improved, thereby improving the overall performance of the coolant performance prediction model.

[0062] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and not limiting of the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can derive other embodiments based on these drawings without inventive efforts.

[0064] Figure 1 A flow chart of a method for detecting the quality of a high-cleanliness and long-life compressor coolant according to an embodiment of the present invention is exemplarily shown;

[0065] Figure 2 A schematic diagram exemplarily shows a coolant performance prediction model according to an embodiment of the present invention;

[0066] Figure 3 A block diagram of a high-cleanliness and long-life compressor coolant quality detection system according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0067] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0068] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0069] Figure 1 A flow chart of a method for detecting the quality of a high-cleanliness, long-life compressor coolant according to an embodiment of the present invention is exemplarily shown. The method comprises:

[0070] Step S1, adding a second preset volume of detergent to multiple barrels of base oil of a first preset volume, respectively, to obtain multiple barrels of first added oil, wherein the detergent includes multiple detergents, and the proportions of the multiple detergents added to each barrel of base oil are different;

[0071] Step S2, sampling each barrel of the first added oil, and performing a sludge control test, a corrosion test, and a deposit test on the first samples obtained by sampling, to obtain a sludge control score, a corrosion resistance score, and a deposit score for each barrel of the first added oil;

[0072] Step S3, sampling each barrel of the first added oil, and detecting the first oxidation induction period score and the first thermogravimetric performance score of each second sample;

[0073] Step S4, dividing each barrel of the first added oil into a plurality of equal portions, adding a third preset volume of an antioxidant to each portion of the first added oil to obtain a plurality of portions of the second added oil, sampling each of the second added oil to obtain third samples, and determining a second oxidation induction period score and a second thermal weight loss performance score for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different;

[0074] Step S5, screening a third added oil from the plurality of second added oils based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score;

[0075] Step S6, dividing each portion of the third added oil into a plurality of portions, adding a fourth preset volume of an anti-wear agent to each portion of the third added oil, to obtain a plurality of portions of fourth added oil, wherein the anti-wear agent includes a plurality of anti-wear substances, and the ratio of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil is different;

[0076] Step S7, determining the anti-wear score of each portion of the fourth added oil and determining the industrial performance index of each portion of the fourth added oil;

[0077] Step S8, obtaining the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score, and industrial performance index of each fourth added oil, and training a coolant performance prediction model to obtain a trained coolant performance prediction model;

[0078] Step S9: Using the trained coolant performance prediction model, simulate tests are performed on the detergent, antioxidant, and anti-wear agent to obtain test results.

[0079] According to the high-cleanliness and long-life compressor coolant quality detection method of an embodiment of the present invention, sludge control test, corrosion test and deposit test can be carried out on the ratios of various detergents, and the improvement of oxidation induction period and thermal weight loss performance of antioxidants can be tested. The anti-wear performance of anti-wear agents can also be tested. During the test process, the preparation process of detergents, antioxidants and anti-wear agents is followed, and a coolant performance prediction model is trained based on the preparation process, so that the performance of coolants with more ratios can be predicted by the coolant performance prediction model, thereby reducing the difficulty and cost of testing and facilitating the search for an optimized ratio.

[0080] According to one embodiment of the present invention, in step S1, polyalphaolefins (PAO), polyesters (POE), polyethers (PAG), and the like can be used as base oils. A predetermined proportion of additives, such as detergents, antioxidants, and antiwear agents, can be added to the base oils to prepare the coolant. Multiple barrels (e.g., 5 or 10 barrels) of a first predetermined volume (e.g., 8 liters) of base oil are first placed, and a second predetermined volume (e.g., 1 liter) of detergent is added to each barrel. The detergents include various detergents, with the proportions of the various detergents in each barrel of base oil varying. The detergents may include sulfonates, alkylphenols, alkylsalicylates, naphthenates, and other detergents, with the proportions of the detergents in each barrel varying. After the detergents are added, the mixture is mixed evenly to produce the first additive oil.

[0081] According to one embodiment of the present invention, in step S2, each barrel of the first added oil may be sampled and the resulting first samples may be tested. Various tests may be performed on the detergent's performance. For example, a sludge control test may be conducted to determine sludge control performance. For example, the sludge weight W1 may be measured using the ASTM D6335 standard test method. This weight may be compared with the average weight W2 of sludge produced when various existing coolants are tested using the same standard to determine a sludge control score. For example, the sludge control score may be calculated as (W2 - W1) / W2. For example, a corrosion test may be conducted to determine corrosion resistance performance. For example, the weight loss W3 of the corroded test sample may be measured using the ASTM D6594 standard test method. A greater weight loss indicates more severe corrosion. This weight loss may be compared with the average weight loss W4 of various existing coolants tested using the same standard to determine a corrosion resistance score. For example, the corrosion resistance score may be calculated as (W4 - W3) / W4. For example, a deposit test can be performed to determine the deposit dispersion performance. For example, the weight W5 of the deposit can be measured using the standard test method of ASTM D7097. This weight can be compared with the average weight W6 of the deposits when multiple existing coolants are tested using the same standard to determine the deposit score. For example, (W6-W5) / W6 can be used as the deposit score.

[0082] According to one embodiment of the present invention, in step S3, to determine the performance of the antioxidant, comparative testing can be performed on the added oil before and after the antioxidant is added. First, before adding the antioxidant, the oxidation induction period and thermal gravimetric performance of the second sample of the first added oil are tested. For example, the first oxidation induction period T1 of each second sample can be measured using the ASTM D3895 standard test method. This can be compared with the average oxidation induction period Ta obtained by testing multiple existing coolants using the same standard to determine a first oxidation induction period score. For example, T1 / Ta can be used as the first oxidation induction period score. For example, parameters such as the onset decomposition temperature, peak decomposition temperature, and residual ash content can be measured using the ASTM E1131 standard test method. These parameters are normalized and weighted summed to obtain a first thermal gravimetric performance parameter H1 of the second sample. This can be compared with the average thermal gravimetric performance parameter Ha obtained by testing multiple existing coolants using the same standard to determine a first thermal gravimetric performance score. For example, H1 / Ha can be used as the first thermal gravimetric performance score.

[0083] According to one embodiment of the present invention, in step S4, to test the synergistic effect of the antioxidant and detergent, each barrel of the first added oil can be divided equally into multiple portions. A third predetermined volume of antioxidant is added to each portion of the first added oil to obtain a second added oil. The antioxidant includes multiple antioxidant substances (e.g., phenolic antioxidant T502, antioxidant preservative T202, etc.), and the ratios of the antioxidant substances added to each portion of the first added oil vary. After obtaining the second added oil, the second added oil can be sampled. The third samples obtained from each portion of the second added oil are then used in the same manner as above to determine a second oxidative induction period score and a second thermal gravimetric performance score for each portion of the second added oil.

[0084] According to one embodiment of the present invention, in step S5, a third added oil is screened out from a plurality of second added oils based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score, including: selecting a candidate second added oil from the plurality of second added oils whose second oxidation induction period score is higher than a preset induction period score lower limit and whose second thermal weight loss performance score is higher than a preset thermal weight loss performance score lower limit; determining an oxidation induction period score increase based on the first oxidation induction period score and the second oxidation induction period score of the second added oil, and determining a thermal weight loss performance increase based on the first thermal weight loss performance score and the second thermal weight loss performance score of the second added oil; taking a weighted sum of the oxidation induction period score increase, the thermal weight loss performance increase sum, the sludge control score, the corrosion protection score, and the deposit score to obtain a synergistic performance score; and screening out the third added oil from the plurality of candidate second added oils based on the synergistic performance score.

[0085] According to one embodiment of the present invention, the preset lower limit of the induction period score for each portion of the second added oil can be set based on the first oxidation induction period score of the corresponding first added oil. As mentioned above, each portion of the second added oil is obtained by dividing a barrel of the first added oil into multiple portions and then adding an antioxidant. Therefore, each barrel of the first added oil corresponds to multiple portions of the second added oil. The preset lower limit of the induction period score for a portion of the second added oil can be equal to the first oxidation induction period score of the corresponding first added oil multiplied by a preset coefficient greater than 1, for example, 1.1. The present invention does not limit the specific value of the preset lower limit of the induction period score. Similarly, a preset lower limit of the thermal weight loss performance score can be set for each portion of the second added oil. Among the multiple portions of the second added oil, a candidate second added oil is selected whose second oxidation induction period score is higher than the preset lower limit of the induction period score and whose second thermal weight loss performance score is higher than the preset lower limit of the thermal weight loss performance score. In other words, a second added oil with a qualified second oxidation induction period score and a qualified second thermal weight loss performance score is selected as the candidate second added oil. This eliminates unqualified second added oils, reduces the number of subsequent experiments, and improves experimental efficiency.

[0086] According to one embodiment of the present invention, for a qualified candidate second added oil, the increase in the oxidation induction period score after the addition of the antioxidant can be determined based on the first oxidation induction period score of the corresponding first added oil and its second oxidation induction period score, that is, the increase in the oxidation induction period score can be determined by the formula (second oxidation induction period score - first oxidation induction period score) / first oxidation induction period score. Similarly, the increase in the thermal weight loss performance after the addition of the antioxidant can be determined based on the first thermal weight loss performance score of the corresponding first added oil and its second thermal weight loss performance score, that is, the increase in the thermal weight loss performance can be determined by the formula (second thermal weight loss performance score - first thermal weight loss performance score) / first thermal weight loss performance score.

[0087] According to one embodiment of the present invention, the above increase in the oxidation induction period score and the increase in the thermal weight loss performance are caused by the antioxidant. Therefore, the above two increases can reflect the performance of the antioxidant, and the sludge control score, the anti-corrosion score and the deposit score can reflect the performance of the detergent. The increase in the oxidation induction period score, the increase in the thermal weight loss performance, the sludge control score, the anti-corrosion score and the deposit score are weighted and summed to obtain the synergistic performance score of the antioxidant and detergent added to the second added oil.

[0088] According to one embodiment of the present invention, a third additive oil can be selected from multiple candidate second additive oils based on their synergistic performance scores. For example, the second additive oil with the highest synergistic performance score of 10 is selected as the third additive oil. The antioxidant and detergent in the third additive oil exhibit a strong synergistic effect, allowing them to simultaneously exert the effects of both the detergent and antioxidant.

[0089] According to one embodiment of the present invention, in step S6, the third additive oil can be determined to have relatively good detergency and antioxidant properties. Further testing can be performed on the antiwear agent to determine which ratio of antiwear agent best improves antiwear performance. Each portion of the third additive oil can be divided into multiple portions, and a fourth preset volume of antiwear agent is added to each portion of the third additive oil to obtain multiple portions of the fourth additive oil. Furthermore, the antiwear agent includes multiple antiwear substances (e.g., lanthanum dialkyldithiophosphate (LaDDP), organic borate (OB), bis(dimethylamino) hydrogenated cardanol phosphate, etc.), and the ratios of the multiple antiwear substances added to each portion of the fourth additive oil vary.

[0090] According to one embodiment of the present invention, in step S7, a four-ball friction test can be used to determine the anti-wear score of each portion of the fourth additive oil. For example, the wear spot diameter R1 of each portion of the fourth additive oil can be determined through the four-ball friction test. Multiple existing coolants can then be tested using the same experimental standards to obtain an average wear spot diameter R2 for each coolant. The wear spot score is then calculated as (R2-R1) / R2. The wear volume V1 of each portion of the fourth additive oil can also be determined through the four-ball friction test. Multiple existing coolants can then be tested using the same experimental standards to obtain an average wear volume V2 for each coolant. The wear volume score is then calculated as (V2-V1) / V2. Furthermore, the wear spot score and the wear volume score can be weighted and summed to obtain the anti-wear score for each portion of the fourth additive oil.

[0091] According to one embodiment of the present invention, various industrial performance indicators can be measured for each portion of the fourth additive oil. For example, if the coolant is relatively viscous, the operating resistance of the equipment being filled with the coolant may increase, thereby increasing power consumption. The increase in the power consumption of the equipment after adding the fourth additive oil relative to the average power consumption of the same equipment after adding multiple existing coolants can be determined as one of the industrial performance indicators. In another example, the pour point of each portion of the fourth additive oil can be measured as one of the industrial performance indicators. Other industrial performance indicators can also be measured, and the present invention does not limit the type of industrial performance indicator.

[0092] According to one embodiment of the present invention, in step S8, the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score, and industrial performance index of each fourth added oil can be obtained. The fourth added oil is derived from the third added oil, and the third added oil is derived from the first added oil. Therefore, the sludge control score, corrosion protection score, and deposit score of the first added oil corresponding to the fourth added oil are the sludge control score, corrosion protection score, and deposit score of the fourth added oil. The second oxidation induction period score and second thermal weight loss performance score of the third added oil corresponding to the fourth added oil are the second oxidation induction period score and second thermal weight loss performance score of the fourth added oil.

[0093] According to one embodiment of the present invention, in step S8, the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score, and industrial performance index of each portion of the fourth added oil are obtained, and a coolant performance prediction model is trained to obtain a trained coolant performance prediction model, including: obtaining the amount and ratio of detergent, the amount and ratio of antioxidant, and the amount and ratio of anti-wear agent in each portion of the fourth added oil; inputting the amount and ratio of detergent into the first coding level; Obtain the detergent addition feature vector; input the amount and ratio of the antioxidant into the second encoding layer to obtain the antioxidant addition feature vector; input the amount and ratio of the antiwear agent into the third encoding layer to obtain the antiwear agent addition feature vector; input the detergent addition feature vector into the first fully connected layer to obtain the predicted sludge control score, the predicted anti-corrosion score and the predicted deposit score; input the detergent addition feature vector and the antioxidant addition feature vector into the first attention mechanism layer to obtain the antioxidant feature vector; input the antioxidant feature vector into the second fully connected layer to obtain the predicted sludge control score, the predicted anti-corrosion score and the predicted deposit score. The method comprises the following steps: connecting the layers to obtain the predicted oxidation induction period score and the predicted thermal weight loss performance score; inputting the antioxidant feature vector and the anti-wear agent addition feature vector into the second attention mechanism layer to obtain the anti-wear feature vector; inputting the anti-wear feature vector into the third fully connected layer to obtain the predicted anti-wear score; inputting the detergent addition feature vector, the antioxidant addition feature vector and the anti-wear agent addition feature vector into the third attention mechanism layer to obtain the industrial performance feature vector; inputting the industrial performance feature vector into the fourth fully connected layer to obtain the predicted industrial performance index; according to the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, the loss function of the coolant performance prediction model is obtained; the coolant performance prediction model is trained according to the loss function of the coolant performance prediction model to obtain a trained coolant performance prediction model.

[0094] Figure 2 A schematic diagram of a coolant performance prediction model according to an embodiment of the present invention is exemplarily shown.

[0095] According to one embodiment of the present invention, the coolant performance prediction model includes a first encoding layer, a second encoding layer, a third encoding layer, a first fully connected layer, a first attention mechanism layer, a second fully connected layer, a second attention mechanism layer, a third fully connected layer, a third attention mechanism layer, and a fourth fully connected layer.

[0096] According to one embodiment of the present invention, as described above, the fourth additional oil is derived from the third additional oil, which is derived from the first additional oil. Therefore, the amount and ratio of the detergent in the fourth additional oil can be determined based on the amount of detergent in the first additional oil and the number of portions of the first additional oil ultimately divided into the fourth additional oil. Furthermore, the amount and ratio of the antioxidant in the fourth additional oil can be determined based on the number of portions of the third additional oil divided into the fourth additional oil, as well as the amount and ratio of the antioxidant in the third additional oil. The amount and ratio of the antiwear agent in the fourth additional oil can also be determined.

[0097] According to one embodiment of the present invention, the amount and ratio of detergent added can be combined into a vector. The multiple data in the vector may include the amount of detergent added and the proportion of each detergent substance. This vector can be input into the first coding layer, which may include multiple linear operation layers and activation layers. The vector can be processed to obtain a detergent addition feature vector. The detergent addition feature vector is used to describe the performance characteristics of the added detergent. The dimension of the detergent addition feature vector can be higher than the vector composed of the amount and ratio of detergent added. Similarly, the second coding layer can process the amount and ratio of antioxidant added to obtain an antioxidant addition feature vector, and the third coding layer can process the amount and ratio of antiwear agent added to obtain an antiwear agent addition feature vector.

[0098] According to one embodiment of the present invention, as described above, when determining the sludge control score, the anti-corrosion score, and the deposit score, the first added oil is modulated based solely on the amount and ratio of the detergent added, and the sludge control score, the anti-corrosion score, and the deposit score are obtained. Therefore, when determining the predicted sludge control score, the predicted anti-corrosion score, and the predicted deposit score, only the detergent addition feature vector corresponding to the amount and ratio of the detergent added may be referred to. The detergent addition feature vector may be processed by the first fully connected layer and then processed by the activation function to obtain the predicted sludge control score, the predicted anti-corrosion score, and the predicted deposit score.

[0099] According to one embodiment of the present invention, as described above, when determining the second oxidative induction period score and the second thermogravimetric performance score, not only the amount and ratio of the antioxidant is considered, but also the amount and ratio of the detergent is considered, thereby determining a synergistic effect between the two. Therefore, when determining the predicted oxidative induction period score and the predicted thermogravimetric performance score, the amount and ratio of the detergent and the amount and ratio of the antioxidant can also be combined for comprehensive processing. Specifically, the detergent addition feature vector and the antioxidant addition feature vector are processed through the first attention mechanism layer to obtain an antioxidant feature vector that reflects the performance of the detergent and antioxidant. This antioxidant feature vector is then input into the second fully connected layer, and after further processing through the activation function, the predicted oxidative induction period score and the predicted thermogravimetric performance score can be obtained.

[0100] According to one embodiment of the present invention, as described above, when determining the antiwear score, the fourth additive oil includes not only an antiwear agent but also a detergent and an antioxidant. With the synergistic effect of these three additives, the antiwear score of the fourth additive oil is determined through testing. Therefore, the amount and ratio of the detergent, the amount and ratio of the antioxidant, and the amount and ratio of the antiwear agent can be combined to determine the predicted antiwear score. As described above, the antioxidant feature vector is a vector obtained by combining the amount and ratio of the detergent and the amount and ratio of the antioxidant. Therefore, the antioxidant feature vector and the antiwear agent addition feature vector can be input into the second attention mechanism layer for processing to obtain an antiwear feature vector that reflects the combined performance of the detergent, antioxidant, and antiwear agent. This antiwear feature vector can then be input into the third fully connected layer, and after further processing with a refined activation function, the predicted antiwear score can be obtained.

[0101] According to one embodiment of the present invention, the detergent addition feature vector, the antioxidant addition feature vector and the anti-wear agent addition feature vector can also be directly processed through the third attention mechanism layer to obtain the industrial performance feature vector. Although the industrial performance index is also a comprehensive performance index obtained by combining the addition amount and ratio of the detergent, the addition amount and ratio of the antioxidant, and the addition amount and ratio of the anti-wear agent, in order to increase the differentiation between the two branches of the industrial performance index and the predicted anti-wear score during training, the industrial performance feature vector can be obtained separately through the third attention mechanism layer and input into the fourth fully connected layer. After processing through the activation function, the predicted industrial performance index can be obtained.

[0102] In this way, the processing flow of the coolant performance prediction model can be made consistent with the test flow of various test indicators of the coolant, so that the coolant performance prediction model can better simulate the test flow of the coolant and improve the structural scientificity and prediction accuracy of the coolant performance prediction model.

[0103] According to one embodiment of the present invention, a loss function of a coolant performance prediction model is obtained based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, including: obtaining a cleanliness performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score and the deposit score; obtaining an antioxidant performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score and the second thermal weight loss performance score. energy loss function; obtain the anti-wear performance loss function according to the predicted sludge control score, predicted anti-corrosion score, sludge control score, anti-corrosion score, predicted deposit score, deposit score, predicted oxidation induction period score, predicted thermal weight loss performance score, second oxidation induction period score, second thermal weight loss performance score, predicted anti-wear score and anti-wear score; obtain the industrial performance loss function according to the predicted sludge control score, predicted anti-corrosion score, predicted deposit score, predicted oxidation induction period score, predicted thermal weight loss performance score, predicted anti-wear score and predicted industrial performance indicators, as well as the sludge control score, anti-corrosion score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score and industrial performance indicators; determine the loss function of the coolant performance prediction model according to the cleaning performance loss function, antioxidant performance loss function, anti-wear performance loss function and industrial performance loss function.

[0104] According to one embodiment of the present invention, a cleaning performance loss function is obtained based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, and the deposit score, including: obtaining the cleaning performance loss function according to formula (1): , (1), in, To predict the sludge control score, Score sludge control, To predict the corrosion resistance score, Score for corrosion resistance, To score the predicted sediment, Score the sediment, 、 and The preset weights.

[0105] According to one embodiment of the present invention, in formula (1), The difference between the sludge control score and the predicted sludge control score is used as the training label, and the difference between the predicted sludge control score output by the model and the training label is used to form a cleaning performance loss function for training. In this way, the predicted sludge control score is closer to the measured value during training, reducing the model error and improving the model accuracy. Similarly, the error between the predicted anti-corrosion score and the anti-corrosion score can also be determined. , predicted sediment scores and sediment score errors , and take the weighted average of the three to obtain the clean performance loss function.

[0106] In this way, the measured values ​​of each score can be used as training labels, and the error between the predicted score output by the model and the training label can be used to form a loss function, so that the model output value is close to the measured value during the training process, thereby improving the model accuracy.

[0107] According to one embodiment of the present invention, the antioxidant performance loss function is obtained based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score and the second thermal weight loss performance score, including: obtaining the antioxidant performance loss function according to formula (2) , (2), in, To predict the oxidative induction period score, Score the second oxidation induction period, To score the predicted thermogravimetric performance, For the second thermal weight loss performance score, and is the preset weight, and max is the maximum value function.

[0108] According to one embodiment of the present invention, in formula (2), is the error between the second oxidative induction period score and the predicted oxidative induction period score, is the error between the second thermogravimetric performance score and the predicted thermogravimetric performance score, It is the weighted average of the errors between the scores output by the model and the training labels.

[0109] According to one embodiment of the present invention, as described above, when obtaining the predicted oxidation induction period score and the predicted thermal weight loss performance score, the first attention mechanism layer is used to process the detergent addition feature vector and the antioxidant addition feature vector. Therefore, the predicted oxidation induction period score and the predicted thermal weight loss performance score are obtained by combining the outputs of the first encoding layer and the second encoding layer. Both the first encoding layer and the second encoding layer may contain errors. Therefore, the maximum value function can be used to select the training branch. If the error of the first encoding layer itself is large, , which can be expressed as The corresponding error may be mainly caused by the error of the first coding level, which can be The first coding layer and the first fully connected layer are trained so that the error when combining the outputs of the first coding layer and the second coding layer is reduced. , it means The corresponding error is not mainly caused by the error of the first coding level, so it can be The branches corresponding to the first encoding layer and the second encoding layer are trained comprehensively without strengthening the training of the first encoding layer alone. That is, the first encoding layer, the second encoding layer, the first attention mechanism layer, and the second fully connected layer are trained.

[0110] According to one embodiment of the present invention, the anti-wear performance loss function is obtained based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, the second thermal weight loss performance score, the predicted anti-wear score and the anti-wear score, including: obtaining the anti-wear performance loss function according to formula (3) , (3), in, To predict the anti-wear score, Rating for wear resistance.

[0111] According to one embodiment of the present invention, is the error between the predicted anti-wear score and the anti-wear score. The meaning of the maximum function is similar to the above, that is, since the predicted anti-wear score is obtained by integrating the outputs of the first coding level, the second coding level, and the third coding level, the maximum function can be used to determine whether the error between the outputs of the first coding level and the second coding level is too large. Is the corresponding error mainly caused by the error of the first coding level and the second coding level? If so, the training of the branch corresponding to the first coding level or the branch corresponding to the second coding level can be strengthened separately. Otherwise, The corresponding error is the largest, then If the corresponding error is not mainly caused by the error of the first coding level and the second coding level, it is not necessary to perform intensive training on the branch corresponding to the first coding level or the branch corresponding to the second coding level. Instead, Just perform comprehensive training, that is, perform comprehensive training on the first encoding layer, the second encoding layer, the third encoding layer, the first attention mechanism layer, the second attention mechanism layer and the third fully connected layer.

[0112] In this way, when determining the antioxidant performance loss function and the anti-wear performance loss function, the main source of error can be determined by taking the maximum value function, so as to judge whether it is necessary to conduct separate intensive training on the previous branches, so as to reduce the errors of each branch while reducing the overall error, thereby improving the targetedness and efficiency of training and improving the training quality.

[0113] According to one embodiment of the present invention, the industrial performance loss function is obtained based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, including: obtaining the industrial performance loss function according to formula (4) , (4), in, is the i-th industrial performance index, is the i-th predicted industrial performance indicator, n is the number of types of industrial performance indicators, i≤n, and both i and n are positive integers.

[0114] According to one embodiment of the present invention, in formula (4), It is the average value of the relative error between the predicted industrial performance index and the industrial performance index. It is a loss function that combines the errors of the first coding level, the second coding level, and the third coding level. When constructing the industrial performance loss function, Set as the numerator, Set as the denominator, during training, not only can By reducing the number of channels, the training of the branches corresponding to the first, second and third coding levels can be strengthened. Expand, so that The first encoding layer, the second encoding layer, the third encoding layer, the third attention mechanism layer and the fourth fully connected layer are trained, and the training of the first encoding layer, the second encoding layer and the third encoding layer can be further strengthened.

[0115] In this way, by constructing the industrial performance loss function, the training of the branches corresponding to the first coding level, the second coding level and the third coding level can be strengthened, and the average value of the relative error between the predicted industrial performance indicators and the industrial performance indicators can be reduced during training to train the first coding level, the second coding level, the third coding level, the third attention mechanism level and the fourth fully connected layer, thereby further strengthening the training of the first coding level, the second coding level and the third coding level, so as to comprehensively improve the accuracy of feature information extraction of the first coding level, the second coding level and the third coding level through multiple training of the first coding level, the second coding level and the third coding level, thereby improving the overall performance of the coolant performance prediction model.

[0116] According to one embodiment of the present invention, after obtaining the above-mentioned cleaning performance loss function, antioxidant performance loss function, anti-wear performance loss function, and industrial performance loss function, a weighted summation of these loss functions can be performed to obtain the loss function of the coolant performance prediction model. During training, backpropagation can be performed based on the loss functions to train the branches corresponding to each loss function, and multiple reinforcement training can be performed on the first, second, and third coding levels to improve the overall performance of the coolant performance prediction model. After multiple training sessions, for example, after processing and training the amount and ratio of the detergent, antioxidant, and anti-wear agent added to each portion of the fourth added oil, a trained coolant performance prediction model can be obtained.

[0117] According to one embodiment of the present invention, in step S9, a simulation test can be performed on the detergent, antioxidant and anti-wear agent based on the trained coolant performance prediction model to obtain test results. For example, the addition amount and ratio of the detergent, antioxidant and anti-wear agent that have not been tested can be input into the trained coolant performance prediction model, and then the predicted industrial performance index (i.e., the simulated test result) can be obtained, which can be used to find the optimized ratio of the detergent, antioxidant and anti-wear agent. For example, after adjusting the ratio of the detergent, antioxidant and anti-wear agent, the trained coolant performance prediction model can be input for simulation testing to obtain the predicted value of the industrial performance index, so that the experimenter can more conveniently determine the relationship between the ratio of the detergent, antioxidant and anti-wear agent and the industrial performance index, which helps the experimenter find the optimized ratio of the detergent, antioxidant and anti-wear agent that meets the requirements. In the process of finding the optimized ratio, the number of actual experiments is reduced, the search efficiency is improved, and the experimental cost is reduced.

[0118] According to an embodiment of the present invention, a high-cleanliness, long-life compressor coolant quality testing method can be used to test various detergent ratios for sludge control, corrosion, and deposits, as well as antioxidants for improving oxidation induction period and thermal weight loss performance. Furthermore, the anti-wear performance of anti-wear agents can be tested. During the testing process, the formulation process of the detergent, antioxidant, and anti-wear agent is followed, and a coolant performance prediction model is trained based on this formulation process. This allows the coolant performance prediction model to predict the performance of a wider range of coolant ratios, thereby reducing testing difficulty and cost and facilitating the search for optimized ratios. When constructing the coolant performance prediction model, the model's processing flow can be aligned with the test flow for various coolant test indicators, enabling the model to better simulate the coolant testing process and improving the model's structural scientificity and prediction accuracy. When determining the cleanliness performance loss function, the measured values ​​of each score are used as training labels, and the error between the predicted scores output by the model and the training labels is used to form the loss function. This ensures that the model output values ​​are close to the measured values ​​during training, improving model accuracy. When determining the antioxidant performance loss function and the anti-wear performance loss function, the main source of error can be determined by taking the maximum function, thereby judging whether it is necessary to conduct separate intensive training on the previous branches, so as to reduce the errors of each branch while reducing the overall error, thereby improving the pertinence and efficiency of training and improving the training quality. It is also possible to strengthen the training of the branches corresponding to the first coding level, the second coding level, and the third coding level by constructing an industrial performance loss function. It is also possible to reduce the average value of the relative error between the predicted industrial performance index and the industrial performance index during training to train the first coding level, the second coding level, the third coding level, the third attention mechanism level, and the fourth fully connected layer, thereby further strengthening the training of the first coding level, the second coding level, and the third coding level. Through multiple training of the first coding level, the second coding level, and the third coding level, the accuracy of feature information extraction of the first coding level, the second coding level, and the third coding level is comprehensively improved, thereby improving the overall performance of the coolant performance prediction model.

[0119] Figure 3 A block diagram of a high-cleanliness and long-life compressor coolant quality detection system according to an embodiment of the present invention is exemplarily shown, wherein the system comprises:

[0120] a first oil addition module for adding a second preset volume of detergent to a plurality of barrels of base oil of a first preset volume, respectively, to obtain a plurality of barrels of first added oil, wherein the detergent comprises a plurality of detergent substances, and the proportions of the plurality of detergent substances in the detergent added to each barrel of base oil are different;

[0121] a cleanliness test module, for sampling each barrel of first-added oil, and performing sludge control test, corrosion test, and sediment test on the first sample obtained, to obtain a sludge control score, corrosion resistance score, and sediment score for each barrel of first-added oil;

[0122] a first oxidation test module, for sampling each barrel of the first added oil and detecting a first oxidation induction period score and a first thermal weight loss performance score of each second sample;

[0123] a second oxidation test module, configured to divide each barrel of the first added oil into a plurality of equal portions, add a third preset volume of an antioxidant to each portion of the first added oil to obtain a plurality of portions of the second added oil, sample the second added oil to obtain third samples, and determine a second oxidation induction period score and a second thermal weight loss performance score for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different;

[0124] a screening module for screening a third added oil from a plurality of second added oils based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score;

[0125] a fourth oil addition module, configured to divide each portion of the third added oil into a plurality of portions, and to add a fourth preset volume of anti-wear agent to each portion of the third added oil to obtain a plurality of fourth added oil portions, wherein the anti-wear agent comprises a plurality of anti-wear substances, and the ratios of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil are different from each other;

[0126] an anti-wear test module for determining an anti-wear score for each portion of the fourth additive oil and determining an industrial performance index for each portion of the fourth additive oil;

[0127] A training module is used to obtain a sludge control score, an anti-corrosion score, a deposit score, a second oxidation induction period score, a second thermal weight loss performance score, an anti-wear score, and an industrial performance index of each fourth added oil, and to train a coolant performance prediction model to obtain a trained coolant performance prediction model;

[0128] The simulation test module is used to simulate the test of detergents, antioxidants and anti-wear agents using the trained coolant performance prediction model to obtain test results.

[0129] The present invention may be a method, an apparatus, a system and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for executing various aspects of the present invention.

[0130] Those skilled in the art will appreciate that the embodiments of the present invention described above and shown in the accompanying drawings are intended to be illustrative only and are not intended to limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and illustrated in the embodiments. Any variations or modifications may be made to the embodiments of the present invention without departing from the principles described.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting the quality of high-cleanliness and long-life compressor coolant, characterized in that: include: adding a second preset volume of detergent to multiple barrels of base oil of a first preset volume, respectively, to obtain multiple barrels of first added oil, wherein the detergent includes multiple detergents, and the proportions of the multiple detergents in each barrel of base oil are different; Sampling each barrel of first-added oil, and subjecting the first samples obtained from the sampling to sludge control test, corrosion test, and sediment test, respectively, to obtain a sludge control score, corrosion resistance score, and sediment score for each barrel of first-added oil; Sampling each barrel of the first added oil, and testing the first oxidation induction period score and the first thermal weight loss performance score of each second sample; Each barrel of the first added oil is divided into a plurality of equal portions, a third preset volume of an antioxidant is added to each portion of the first added oil to obtain a plurality of portions of the second added oil, and the second added oil is sampled to obtain third samples, and a second oxidation induction period score and a second thermal weight loss performance score are determined for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different from each other; Screening a third added oil from a plurality of second added oils according to the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score; dividing each portion of the third added oil into a plurality of portions, and adding a fourth preset volume of an anti-wear agent to each portion of the third added oil to obtain a plurality of portions of the fourth added oil, wherein the anti-wear agent comprises a plurality of anti-wear substances, and the ratios of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil are different from each other; Determining an anti-wear score for each fourth added oil and determining an industrial performance index for each fourth added oil; Obtaining a sludge control score, an anti-corrosion score, a deposit score, a second oxidation induction period score, a second thermal weight loss performance score, an anti-wear score, and an industrial performance index of each fourth added oil, and training a coolant performance prediction model to obtain a trained coolant performance prediction model; Using the trained coolant performance prediction model, simulated tests were conducted on detergents, antioxidants, and anti-wear agents to obtain test results.

2. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 1, characterized in that: According to the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score, a third added oil is screened out from multiple second added oils, including: Among the plurality of second added oils, selecting a candidate second added oil having a second oxidation induction period score higher than a preset induction period score lower limit and a second thermal weight loss performance score higher than a preset thermal weight loss performance score lower limit; Determining an oxidation induction period score increase based on the first oxidation induction period score and the second oxidation induction period score of the second added oil, and determining a thermal weight loss performance increase based on the first thermal weight loss performance score and the second thermal weight loss performance score of the second added oil; The synergistic performance score is obtained by weighted summing the increase in the oxidation induction period score, the increase in the thermal weight loss performance, the sludge control score, the corrosion resistance score, and the deposit score. According to the synergistic performance score, the third added oil is selected from multiple candidate second added oils.

3. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 1, characterized in that: Obtain the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score and industrial performance index of each fourth added oil, and train a coolant performance prediction model to obtain a trained coolant performance prediction model, including: Obtaining the amount and ratio of detergent, antioxidant, and antiwear agent in each portion of the fourth added oil; Inputting the amount and ratio of detergent added into the first coding level to obtain a detergent addition feature vector; Inputting the amount and ratio of the antioxidant into the second coding level to obtain the antioxidant addition feature vector; Input the amount and ratio of the anti-wear agent into the third coding level to obtain the anti-wear agent addition feature vector; Input the detergent addition feature vector into the first fully connected layer to obtain the predicted sludge control score, the predicted corrosion protection score, and the predicted deposit score; Input the detergent addition feature vector and the antioxidant addition feature vector into the first attention mechanism layer to obtain the antioxidant feature vector; The antioxidant feature vector is input into the second fully connected layer to obtain the predicted oxidation induction period score and the predicted thermal weight loss performance score; Input the antioxidant feature vector and the anti-wear agent addition feature vector into the second attention mechanism layer to obtain the anti-wear feature vector; The anti-wear feature vector is input into the third fully connected layer to obtain the predicted anti-wear score; Input the detergent addition feature vector, antioxidant addition feature vector, and antiwear agent addition feature vector into the third attention mechanism layer to obtain the industrial performance feature vector; Input the industrial performance feature vector into the fourth fully connected layer to obtain the predicted industrial performance index; Obtaining a loss function of a coolant performance prediction model based on the predicted sludge control score, the predicted corrosion protection score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score, and the predicted industrial performance index, as well as the sludge control score, the corrosion protection score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score, and the industrial performance index; The coolant performance prediction model is trained according to the loss function of the coolant performance prediction model to obtain a trained coolant performance prediction model.

4. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 3, characterized in that: According to the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score and the industrial performance index, the loss function of the coolant performance prediction model is obtained, including: Obtaining a detergency loss function based on the predicted sludge control score, the predicted corrosion protection score, the sludge control score, the corrosion protection score, the predicted deposit score, and the deposit score; Obtaining an antioxidant performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, and the second thermal weight loss performance score; Obtaining an anti-wear performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, the second thermal weight loss performance score, the predicted anti-wear score, and the anti-wear score; Obtaining an industrial performance loss function based on the predicted sludge control score, the predicted anti-corrosion score, the predicted deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the predicted anti-wear score, and the predicted industrial performance index, as well as the sludge control score, the anti-corrosion score, the deposit score, the second oxidation induction period score, the second thermal weight loss performance score, the anti-wear score, and the industrial performance index; The loss function of the coolant performance prediction model is determined based on the cleaning performance loss function, the anti-oxidation performance loss function, the anti-wear performance loss function and the industrial performance loss function.

5. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 4, characterized in that: Based on the predicted sludge control score, the predicted corrosion protection score, the sludge control score, the corrosion protection score, the predicted deposit score, and the deposit score, a detergency loss function is obtained, including: According to the formula ， Obtaining a clean performance loss function ,in, To predict the sludge control score, Score sludge control, To predict the corrosion resistance score, Score for corrosion resistance, To score the predicted sediment, Score the sediment, 、 and The preset weights.

6. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 5, characterized in that: According to the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score and the second thermal weight loss performance score, the antioxidant performance loss function is obtained, including: According to the formula ， Obtaining antioxidant performance loss function ,in, To predict the oxidative induction period score, Score the second oxidation induction period, To score the predicted thermogravimetric performance, For the second thermal weight loss performance score, and is the preset weight, and max is the maximum value function.

7. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 6, characterized in that: According to the predicted sludge control score, the predicted anti-corrosion score, the sludge control score, the anti-corrosion score, the predicted deposit score, the deposit score, the predicted oxidation induction period score, the predicted thermal weight loss performance score, the second oxidation induction period score, the second thermal weight loss performance score, the predicted anti-wear score and the anti-wear score, the anti-wear performance loss function is obtained, including: According to the formula ， Obtain anti-wear performance loss function ,in, To predict the anti-wear score, Rating for wear resistance.

8. The method for detecting the quality of high-cleanliness and long-life compressor coolant according to claim 7, characterized in that: According to the predicted sludge control score, predicted corrosion protection score, predicted deposit score, predicted oxidation induction period score, predicted thermal weight loss performance score, predicted anti-wear score and predicted industrial performance index, as well as the sludge control score, corrosion protection score, deposit score, second oxidation induction period score, second thermal weight loss performance score, anti-wear score and industrial performance index, the industrial performance loss function is obtained, including: According to the formula ， Obtaining industrial performance loss function ,in, is the i-th industrial performance index, is the i-th predicted industrial performance indicator, n is the number of types of industrial performance indicators, i≤n, and both i and n are positive integers.

9. A high-cleanliness and long-life compressor coolant quality detection system, used to perform the method according to any one of claims 1 to 8, characterized in that: include: a first oil addition module for adding a second preset volume of detergent to a plurality of barrels of base oil of a first preset volume, respectively, to obtain a plurality of barrels of first added oil, wherein the detergent comprises a plurality of detergent substances, and the proportions of the plurality of detergent substances in the detergent added to each barrel of base oil are different; a cleanliness test module, for sampling each barrel of first-added oil, and performing sludge control test, corrosion test, and sediment test on the first sample obtained, to obtain a sludge control score, corrosion resistance score, and sediment score for each barrel of first-added oil; a first oxidation test module, for sampling each barrel of the first added oil and detecting a first oxidation induction period score and a first thermal weight loss performance score of each second sample; a second oxidation test module, configured to divide each barrel of the first added oil into a plurality of equal portions, add a third preset volume of an antioxidant to each portion of the first added oil to obtain a plurality of portions of the second added oil, sample the second added oil to obtain third samples, and determine a second oxidation induction period score and a second thermal weight loss performance score for each of the third samples, wherein the antioxidant comprises a plurality of antioxidant substances, and the ratios of the plurality of antioxidant substances added to each portion of the first added oil are different; a screening module for screening a third added oil from a plurality of second added oils based on the first oxidation induction period score, the second oxidation induction period score, the first thermal weight loss performance score, and the second thermal weight loss performance score; a fourth oil addition module, configured to divide each portion of the third added oil into a plurality of portions, and to add a fourth preset volume of anti-wear agent to each portion of the third added oil to obtain a plurality of fourth added oil portions, wherein the anti-wear agent comprises a plurality of anti-wear substances, and the ratios of the plurality of anti-wear substances in the anti-wear agent added to each portion of the fourth added oil are different from each other; an anti-wear test module for determining an anti-wear score for each portion of the fourth additive oil and determining an industrial performance index for each portion of the fourth additive oil; A training module is used to obtain a sludge control score, an anti-corrosion score, a deposit score, a second oxidation induction period score, a second thermal weight loss performance score, an anti-wear score, and an industrial performance index of each fourth added oil, and to train a coolant performance prediction model to obtain a trained coolant performance prediction model; The simulation test module is used to simulate the test of detergents, antioxidants and anti-wear agents using the trained coolant performance prediction model to obtain test results.

Images