A chromatograph cluster cooperative control method and system for a pesticide detection service

Abstract

The present application relates to the technical field of pesticide detection service, and provides a chromatograph cluster cooperative control method and system for pesticide detection service, the method comprising: obtaining an emergency index of the pesticide to be detected; monitoring the state parameters of each chromatograph in real time, and obtaining the comprehensive availability of the chromatograph according to the state parameters; obtaining a target function according to the emergency index and the comprehensive availability; and realizing the cluster cooperative control of the chromatograph according to the target function. The present application can dynamically allocate the pesticide detection task (sorted according to the emergency index) to the most suitable chromatograph (i.e. according to the comprehensive availability), realize the maximization of the global detection efficiency, break through the detection strategy of the traditional chromatograph cluster scheduling which depends on the "first-come-first-detection" or fixed cycle, and can quickly respond to the sudden pesticide safety event.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of pesticide detection service, in particular to a chromatograph cluster collaborative control method and system for pesticide detection service. BACKGROUND

[0002] Pesticide residue detection is a core link to ensure food safety, and the current mainstream technology relies on chromatographic detection equipment (such as high-performance liquid chromatograph, gas chromatograph-mass spectrometer, etc.). In some detection scenarios, because there are many samples to be detected, a larger number of sample analysis can be performed by setting multiple online chromatographs to form a chromatograph cluster. However, the traditional chromatograph cluster scheduling relies more on the "first come first detection" or fixed cycle strategy, and cannot respond to sudden pesticide safety events, for example: 1. When a pesticide safety problem exposed by network public opinion (such as a certain brand of fruit and vegetable pesticide residue exceeding the standard event) needs to be detected urgently, it lacks a dynamic priority mechanism driven by heat; 2. The detection demand of high-toxicity pesticides (such as organochlorine) and low-residue threshold often causes regulatory lag due to queuing delay. SUMMARY

[0003] In order to solve the problems existing in the traditional chromatograph cluster scheduling, the present application provides a chromatograph cluster collaborative control method and system for pesticide detection service, and the specific technical solutions are as follows:

[0004] A chromatograph cluster collaborative control method for pesticide detection service includes the following steps:

[0005] Obtaining an emergency index of the pesticide to be detected;

[0006] Real-time monitoring of the state parameters of each chromatograph, and obtaining the comprehensive availability of the chromatograph according to the state parameters;

[0007] Obtaining a target function according to the emergency index and the comprehensive availability;

[0008] Realizing the cluster collaborative control of the chromatograph according to the target function.

[0009] The chromatograph cluster collaborative control method for pesticide detection service obtains the emergency index and the comprehensive availability, obtains the target function according to the emergency index and the comprehensive availability, and finally realizes the cluster collaborative control of the chromatograph based on the target function. It can dynamically allocate pesticide detection tasks (sorted according to the emergency index) to the most suitable chromatograph (i.e. according to the comprehensive availability), maximize the global detection efficiency, and break through the detection strategy of the traditional chromatograph cluster scheduling relying on "first come first detection" or fixed cycle. It can quickly respond to sudden pesticide safety events.

[0010] Preferably, the specific method for obtaining the target function includes:

[0011] According to the urgency index and the comprehensive availability, an efficiency item used to represent the matching benefit between the pesticide detection task and the chromatograph is obtained;

[0012] A preset load balancing penalty factor and an ideal average queue length are obtained, and a load penalty item is obtained according to the load balancing penalty factor, the ideal average queue length, the total number of available chromatographs and the total number of pesticides to be detected;

[0013] The target function is obtained according to the efficiency item and the load penalty item.

[0014] Preferably, the specific method for obtaining the urgency index of the pesticide to be detected comprises:

[0015] The network popularity, the self-characteristics and the seasonal risk coefficient of the pesticide to be detected are obtained;

[0016] A network popularity factor item is obtained according to the network popularity of the pesticide to be detected;

[0017] A toxicity-economic impact factor item is obtained according to the toxicity grade, the crop economic value coefficient and the residue limit standard;

[0018] The urgency index is obtained according to the network popularity factor item, the toxicity-economic impact factor item and the seasonal risk coefficient;

[0019] The self-characteristics include the toxicity grade, the corresponding crop economic value coefficient and the residue limit standard of the pesticide to be detected.

[0020] Preferably, the specific method for obtaining the comprehensive availability of the chromatograph comprises:

[0021] The continuous running duration is obtained, and a running duration item used to reflect the fatigue degree of the chromatograph is obtained according to the continuous running duration;

[0022] The remaining life and the theoretical life of the current chromatographic column are obtained, and a remaining life item used to reflect the aging degree of the chromatographic column is obtained according to the remaining life and the theoretical life;

[0023] The queue length to be detected is obtained, and a queue pressure index item used to reflect the task queue pressure of the chromatograph is obtained according to the queue length to be detected;

[0024] The comprehensive availability is obtained according to the running duration item, the remaining life item and the queue pressure index item;

[0025] The state parameters include the continuous running duration, the remaining life and the queue length to be detected.

[0026] Preferably, the target function is expressed as

[0027] Wherein, E and L respectively represent the efficiency item and the load penalty item, Ei represents an urgency index of the pesticide i to be detected, Q j represents the comprehensive availability of the chromatograph j, x ij is a binary decision variable and x ij = 1 indicates that the task i is assigned to the chromatograph j, μ represents a load balancing penalty factor, represents an ideal average queue length, N and M represent the total number of available chromatographs and the total number of pesticides to be detected, respectively.

[0028] Preferably, the urgency index is expressed as

[0029] wherein, respectively represent the network popularity factor term and the toxicity-economic impact factor term, and α1, α2, α3 represent the weight coefficients of the network popularity factor term, the toxicity-economic impact factor term, and the seasonal risk coefficient, respectively, H i represents the network popularity of the pesticide i to be detected, H max represents a preset reference value of the popularity of the pesticide i to be detected, T i represents the toxicity grade of the pesticide i to be detected, C i represents the crop economic value coefficient of the pesticide i to be detected, R i represents the residual limit standard of the pesticide i to be detected, S i represents the seasonal risk coefficient of the pesticide i to be detected.

[0030] Preferably, the comprehensive availability is expressed as

[0031] wherein, respectively represent the runtime length term, the remaining life term, and the queue pressure index term, and w1, w2, w3 represent the weight coefficients of the runtime length term, the remaining life term, and the queue pressure index term, respectively, t j , t max respectively represent the continuous runtime length and the maximum safe runtime limit of the chromatograph j, A j , A new respectively represent the remaining life and the theoretical life of the current chromatographic column of the chromatograph j, d j represents the queue length of the chromatograph j to be detected, e represents the natural constant, and λ represents the queue pressure decay coefficient.

[0032] A chromatograph cluster collaborative control system of a pesticide detection service, for implementing the chromatograph cluster collaborative control method of the pesticide detection service, comprising:

[0033] An urgency index acquisition module for acquiring the urgency index of the pesticide to be detected;

[0034] The comprehensive availability obtaining module is configured to monitor state parameters of each chromatograph in real time and obtain comprehensive availability of the chromatograph according to the state parameters.

[0035] The target function obtaining module is configured to obtain a target function according to the urgency index and the comprehensive availability.

[0036] The cluster collaborative control module is configured to realize cluster collaborative control of the chromatographs according to the target function.

[0037] Preferably, the target function obtaining module comprises:

[0038] The benefit item obtaining unit is configured to obtain an efficiency item for representing matching benefits between the pesticide detection task and the chromatograph according to the urgency index and the comprehensive availability.

[0039] The penalty item obtaining unit is configured to obtain a preset load balancing penalty factor and an ideal average queue length, and obtain a load penalty item according to the load balancing penalty factor, the ideal average queue length, a total number of available chromatographs and a total number of pesticides to be detected.

[0040] The target function obtaining unit is configured to obtain a target function according to the efficiency item and the load penalty item.

[0041] Preferably, the urgency index obtaining module comprises:

[0042] The first obtaining unit is configured to obtain network popularity, self characteristics and a seasonal risk coefficient of the pesticide to be detected.

[0043] The second obtaining unit is configured to obtain a network popularity factor item according to the network popularity of the pesticide to be detected.

[0044] The third obtaining unit is configured to obtain a toxicity-economic impact factor item according to a toxicity grade, a crop economic value coefficient and a residue limit standard.

[0045] The fourth obtaining unit is configured to obtain the urgency index according to the network popularity factor item, the toxicity-economic impact factor item and the seasonal risk coefficient.

[0046] The self characteristics comprise a toxicity grade, a corresponding crop economic value coefficient and a residue limit standard of the pesticide to be detected. BRIEF DESCRIPTION OF DRAWINGS

[0047] The application can be further understood from the following description in conjunction with the accompanying drawings.

[0048] Figure 1 is a schematic diagram of a whole process of a chromatograph cluster collaborative control method of a pesticide detection service in an embodiment of the application;

[0049] Figure 2is a flowchart of a specific method for obtaining an emergency index of a pesticide to be detected in an embodiment of the present application;

[0050] Figure 3 is a flowchart of a specific method for obtaining a comprehensive availability of a chromatograph in an embodiment of the present application;

[0051] Figure 4 is a flowchart of a specific method for obtaining a target function in an embodiment of the present application;

[0052] Figure 5 is a schematic diagram of the overall structure of a chromatograph cluster cooperative control system of a pesticide detection service in an embodiment of the present application;

[0053] Figure 6 is a schematic diagram of the continuous running time of 20 chromatographs in an embodiment of the present application;

[0054] Figure 7 is a schematic diagram of the life status of the chromatographic column of 20 chromatographs in an embodiment of the present application;

[0055] Figure 8 is a schematic diagram of the queue length of 20 chromatographs to be detected in an embodiment of the present application;

[0056] Figure 9 is a comparative schematic diagram of the emergency index of different pesticides to be detected in an embodiment of the present application;

[0057] Figure 10 is a comparative schematic diagram of the comprehensive availability of different chromatographs in an embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical scheme and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with embodiments thereof. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the protection scope of the present application.

[0059] Before explaining the embodiments of the present application, a brief introduction to the prior art will be given.

[0060] At present, there are many pesticide residue detection methods, mainly including biochemical determination method and chromatographic detection method. Among them, the enzyme inhibition rate method in the biochemical determination method is listed as the national recommended standard method due to its characteristics of rapidness, sensitivity, simple operation, low cost, etc., and is widely used in the rapid qualitative preliminary screening detection of organophosphorus and carbamate pesticide residues in fruits and vegetables.

[0061] Chromatography is one of the more accurate detection methods, mainly including gas chromatography, liquid chromatography and gas chromatography-mass spectrometry. Gas chromatography-mass spectrometry is one of the most advanced and accurate detection techniques, which can quickly and accurately analyze all organic pollutants in the sample and perform quantitative analysis. This method has been widely used in pesticide residue detection of fruits and vegetables, and provides reliable technical support for the confirmation of pesticide residues.

[0062] Pesticide residue detection is the core link to ensure food safety. The current mainstream technology relies on chromatographic detection equipment (such as high-performance liquid chromatograph, gas chromatograph-mass spectrometer, etc.). However, the existing technical system has the following defects:

[0063] First, the detection task priority is rigid

[0064] Traditional chromatograph cluster scheduling relies more on "first come first detection" or fixed cycle strategy, which cannot respond to sudden pesticide safety events. For example: when the network public opinion exposes the pesticide safety problem (such as the pesticide residue of some brand of fruits and vegetables exceeds the standard event) needs to be detected urgently, there is no dynamic priority mechanism driven by heat; The detection demand of high toxicity pesticide (such as organochlorine) and low residual threshold is often delayed due to queuing delay, resulting in lag of supervision.

[0065] Second, the utilization rate of equipment resources is unbalanced

[0066] The chromatograph cluster has low coordination efficiency, which is manifested in that the device state (such as the service life of chromatographic column, continuous working time) is not included in the scheduling decision, which easily causes device overload damage or interruption of detection in maintenance, for example, the service life of chromatographic column is accelerated in high temperature environment, but the system has no early warning mechanism; In multi-task scene, the load is uneven, and part of the equipment is on standby while the queue of high-value equipment is accumulated.

[0067] Third, multi-source information is fragmented decision

[0068] Data dimension is missing: the existing technology only focuses on the physicochemical properties of pesticides (such as residual standards), ignoring social attributes such as network public opinion and seasonal risks.

[0069] One of the purposes of the present application is to solve the problem that the traditional chromatograph cluster scheduling relies more on "first come first detection" or fixed cycle strategy, which cannot respond to sudden pesticide safety events, such as Figure 1 As shown in the figure, an embodiment of the present application provides a chromatograph cluster collaborative control method for pesticide detection service, which includes the following steps:

[0070] S1, obtaining the emergency index of the pesticide to be detected.

[0071] The emergency index can be obtained based on public opinion data of the pesticide to be detected, such as first obtaining keywords of the pesticide to be detected, then obtaining real-time network heat value of the pesticide to be detected according to the keywords, and finally obtaining the emergency index based on the real-time network heat value. Specifically, the real-time network heat value can be calculated by the social media discussion amount (such as microblog topic index) and / or search engine query amount (such as Baidu index) of the pesticide to be detected.

[0072] As a preferred technical solution, as shown in step S1, the specific method for obtaining the emergency index of the pesticide to be detected includes: Figure 2

[0073] S11, obtaining the network heat, the self-characteristics and the seasonal risk coefficient of the pesticide to be detected; wherein the self-characteristics include but are not limited to the toxicity grade, the corresponding crop economic value coefficient and the residual limit standard of the pesticide to be detected.

[0074] It can be understood that the network heat of the pesticide to be detected can be calculated based on the social media discussion amount (such as microblog topic index) and / or search engine query amount (such as Baidu index) of at least one keyword corresponding to the pesticide to be detected. When the keyword includes multiple keywords, the network heat can be set as the weighted sum of the social media discussion amount (such as microblog topic index) and / or the search engine query amount (such as Baidu index) of the keyword, or the weighted sum of the social media discussion amount (such as microblog topic index) and the search engine query amount (such as Baidu index). If the pesticide to be detected is exposed to a safety event, the network heat will surge, which can trigger emergency detection.

[0075] Specifically, the network heat of the pesticide to be detected is calculated by the search amount change rate of the keyword of the pesticide to be detected, the discussion frequency and emotional tendency of the related topic, and the reporting density of the mainstream media on the related event.

[0076] Considering that some people may create false heat by abnormal brushing and evaluation control, such as when some sudden pesticide safety accidents occur, competitors of the target user may create false heat by abnormal brushing and evaluation control. In some cases, there is a certain subjective emotional bias for network discussion points related to the network heat of the pesticide to be detected, especially negative public opinion. The existence of the above factors will affect the accuracy of the network heat of the pesticide to be detected.

[0077] In order to improve the accuracy of the network heat, as a preferred technical solution, the specific method for obtaining the network heat of the pesticide to be detected includes the following steps:

[0078] Firstly, obtain the basic data heat of multi-source data covering the social media discussion amount, the search engine index, the e-commerce platform search amount and the news media exposure amount related to the pesticide to be detected. ​

[0079] Secondly, a BERT sentiment analysis model is constructed, and a public opinion sentiment correction factor is obtained based on a public opinion positive and negative tendency value calculated by the BERT sentiment analysis model, so as to enhance the heat in a positive sentiment and suppress the heat in a negative sentiment, and prevent false hype.

[0080] Thirdly, an LSTM model is constructed, and a user credibility factor is obtained by analyzing a user behavior sequence by the LSTM model, so as to evaluate the credibility of a user device fingerprint and identify batch registration accounts to prevent false interactions.

[0081] Fourthly, an anomaly flow is detected based on an isolation forest algorithm, and an anti-cheating factor is obtained.

[0082] Fifthly, a network heat of the pesticide to be detected is obtained by fusing the basic data heat, the public opinion sentiment correction factor, the user credibility factor and the anti-cheating factor.

[0083] It should be noted that the user credibility factor and the anti-cheating factor have the effect of preventing false interaction hype, and can suppress false heat and improve the accuracy of network heat. As for the function formula for calculating the network heat of the pesticide to be detected based on the basic data heat, the public opinion sentiment correction factor, the user credibility factor and the anti-cheating factor, it can be a polynomial, a segmented function or a weighted sum of the basic data heat, the public opinion sentiment correction factor, the user credibility factor and the anti-cheating factor, which can be adjusted according to different scenes in actual application.

[0084] Here, the following function is provided to facilitate obtaining the network heat of the pesticide to be detected based on the basic data heat, the public opinion sentiment correction factor, the user credibility factor and the anti-cheating factor:

[0085]

[0086] wherein H base , EC i , C u and A represent the basic data heat, the public opinion sentiment correction factor, the user credibility factor and the anti-cheating factor respectively.

[0087] The basic data heat can be calculated by the formula wherein D k represents an original value of the kth data source, such as social media discussion volume, search engine index, e-commerce platform search volume or news media exposure volume, D max,k represents a preset maximum value of the kth data source, and W kThe weight coefficient of the kth data source is set according to experience, for example, the weight coefficient of the social media discussion volume is set to 0.4, the weight coefficient of the search engine index is set to 0.3, the weight coefficient of the e-commerce platform search volume is set to 0.2, and the weight coefficient of the news media exposure volume is set to 0.1. μ represents an index factor of the data source, which is not less than 1, and is generally set to a value range of 1.5-2.0, preferably 1.5, which serves to enhance the sensitivity of high activity data and avoid the long tail effect.

[0088] Opinion sentiment correction factor wherein, pos m , pos n respectively represent the sentiment intensity weight of the mth positive sentiment word and the sentiment intensity weight of the nth negative sentiment word, which are output by the BERT sentiment analysis model. Specifically, the BERT sentiment analysis model analyzes the context semantics through the Transformer encoder and outputs the polarity probability of each sentiment word (for example, "high efficiency"→positive probability 0.92, "residual"→negative probability 0.85). Exemplarily, in the sentence "the pesticide has significant insecticidal effect, but there is a residual risk", the BERT sentiment analysis model identifies "significant" as a positive word and outputs pos m =0.88, and "residual risk" as a negative word. The output pos n =0.91. TF-IDF is used to reflect the importance of word frequency, and TF-IDF m , TF-IDF n respectively represent the importance weight of the mth positive sentiment word in the text and the importance weight of the nth negative sentiment word in the text, and M', N' respectively represent the total number of positive sentiment words and negative sentiment words identified in the current text. M'+N'+1 is mainly used to normalize the sentiment score to avoid the virtual high score of long text due to the large number of sentiment words, and also can limit the excessive influence of single extreme comment on the overall sentiment score (such as a large number of abnormal reviews).

[0089] User credibility factor wherein, E h , R h respectively represent the user historical interaction event authenticity score and the behavior correlation degree, H represents the user behavior sequence length, i.e. the LSTM input window size, which determines the model's ability to capture long-term dependencies. F represents the current device fingerprint credibility. F avg represents the historical mean value of the device fingerprint credibility, which serves as a reference value, and the ratio quantifies the abnormality degree of the current device. The tanh function is used to compress the device credibility ratio to the interval [-1, 1] to enhance the sensitivity of the model to extreme values.

[0090] Specifically, LSTM models time series, identifies abnormal behavior patterns (e.g. mutations, high-frequency repeated operations), outputs authenticity probability, and obtains authenticity score according to authenticity probability. LSTM captures the time series dependency of behavior sequence, quantifies the correlation strength of current behavior and historical behavior to obtain the behavior correlation degree, and reflects the consistency of user interest. For example, if a user continuously searches for "pesticide application guide" and "efficient pesticide purchase", the behavior correlation degree can be set to 0.8 (high correlation). It can be understood as a sequence length normalization factor, which mainly avoids numerical inflation caused by long sequences and enhances the comparability of different user behavior lengths. The current device fingerprint credibility can be understood as a credibility score based on device features (IP, browser, hardware information).

[0091] Here, the key role of the LSTM network in the user credibility factor function is briefly described. First, LSTM remembers long-term behavior patterns of users through cell state and gating mechanism (input gate, forget gate, output gate), such as input gate determining whether to retain current behavior (e.g. normal browsing vs. abnormal clicking), forget gate for clearing obsolete behavior records (e.g. low-frequency operations 3 months ago), and output gate generating authenticity score of user historical interaction events. Second, LSTM uses attention mechanism (e.g. time-space attention) to identify key behavior nodes, such as using time attention to focus on peak periods in user behavior sequence (e.g. concentrated purchase during promotion), and using spatial attention to associate different behavior types (e.g. search "pesticide" and browse product detail page).

[0092] E h ·R h It can be understood as giving higher weight to behaviors with high credibility and strong correlation, and avoiding long tail effect. E h , R h output by LSTM are summed by weighting to reflect the overall credibility of user behavior, and combined with the current device fingerprint credibility, to finally generate user credibility factor C u .

[0093] In summary, the user credibility factor function analyzes user behavior sequence through LSTM neural network, comprehensively evaluates the authenticity, correlation and device credibility of user behavior, and finally generates a dynamic user credibility factor, which can be used to identify bulk registered accounts and reduce the interference of credible users on popularity.

[0094] Anti-cheating factor wherein V trepresents the flow characteristic value in the time period t, such as flow rate, packet quantity, etc. σ represents the standard deviation of the flow characteristic value, used to measure the data fluctuation range, and generally, an abnormal threshold is set by 3σ, and data exceeding 3σ of the mean value can be identified as abnormal. T represents the total number of time windows, used to normalize the abnormal frequency, for example, 60 time windows (1 window per minute) in the past 1 hour are counted, I is an indicator function, taking 1 when V t > 3σ, otherwise 0, which is used to mark whether the current time window detects flow anomaly, λ' represents the preset decay coefficient, used to control the decay speed of the influence of historical abnormal events on the current score. t represents the time window number, which is combined with the exponential term e -λ'·t to realize the time decay effect.

[0095] The anti-cheating factor function is analyzed below to understand its meaning:

[0096] represents the proportion of historical time windows that are judged to be abnormal, reflecting the overall frequency of abnormal occurrence, i.e., used to calculate the abnormal frequency. The exponential term e -λ'·t is used to exponentially decay the influence of historical abnormal events, so that the weight of recent abnormalities is higher, and the influence of long-term abnormalities gradually weakens.

[0097] is used to convert the abnormal frequency into a score, and the value closer to 1 indicates a greater likelihood of current flow abnormality. For parameter settings in the anti-cheating factor function, such as the standard deviation σ, it can be adjusted according to the historical data distribution, for example, using a sliding window to calculate the dynamic standard deviation, λ' needs to balance real-time and stability, a value that is too large ignores historical information, and a value that is too small has a lagging response, and the total number of time windows needs to match the business scenario, a short window (such as T=10) is suitable for sudden abnormality, and a long window (such as T=100) is suitable for continuous abnormality detection.

[0098] The anti-cheating factor function is based on the Isolation Forest algorithm to identify abnormal periods in flow data (such as sudden surge behavior), which functions to reduce the value of the anti-cheating factor A when detecting abnormal surge behavior (such as concentrated period explosive growth), thereby suppressing false heat.

[0099] Overall, The numerator part of f (t) fuses the basic heat and sentiment correction, amplifies the heat in positive sentiment, and suppresses the heat in negative sentiment, and the denominator part is normalized to avoid unstable model due to large numerical values; this function fuses sentiment analysis and anti-cheating mechanism, can eliminate false traffic interference and correct public opinion sentiment bias, and adjusts network heat in real time.

[0100] The seasonal risk coefficient range is generally set to 0-1, which can be set by the technical personnel according to experience. In the 30 days before the mature period, such as the mature period of litchi, the seasonal risk coefficient can be set to 0.8; in the non-production season, the seasonal risk coefficient can be set to 0.1. Preferably, a crop growth stage prediction model related to the pesticide to be detected can be constructed through meteorological data (temperature, precipitation) and soil sensor data to identify the sensitive period of pesticide use (such as flowering period, fruit expansion period), different seasonal risk coefficients are assigned to different crop growth stages, and the seasonal risk coefficient is appropriately increased in the sensitive period of pesticide use.

[0101] S12, obtaining a network heat factor term according to the network heat of the pesticide to be detected.

[0102] Suppose H i represents the network heat of the pesticide to be detected i, H max represents the preset heat reference value of the pesticide to be detected i or the preset maximum heat value, then the network heat factor term can be represented as

[0103] The preset heat reference value of the pesticide to be detected i can take the historical peak value or industry threshold (such as public opinion early warning red line), which is mainly used for normalization processing.

[0104] S13, obtaining a toxicity-economic impact factor term according to the toxicity grade, the crop economic value coefficient and the residue limit standard.

[0105] Specifically, suppose T i represents the toxicity grade of the pesticide to be detected i, C i represents the crop economic value coefficient of the pesticide to be detected i, R i represents the residue limit standard of the pesticide to be detected i, then the toxicity-economic impact factor term can be represented as

[0106] The toxicity grade can be divided into five levels of highly toxic, high toxicity, medium toxicity, low toxicity and low toxicity according to the degree of harm, or reference is made to the WHO acute toxicity classification, the higher the toxicity grade, the larger the corresponding T i . Exemplarily, the toxicity grades corresponding to the five levels of highly toxic, high toxicity, medium toxicity, low toxicity and low toxicity can be set to 5, 4, 3, 2 and 1 respectively. The crop economic value coefficient range is generally set to 0-1, such as high value-added crops such as sunshine grape, which can be set to 1.0, bulk crops such as rice can be set to 0.6 or 0.7, etc. In general, the higher the economic value of the crop, the greater the social impact of pesticide residue risk, and the larger the crop economic value coefficient.

[0107] The residual limit standard can be selected as a national residual limit standard or an international residual limit standard, and the lower the value, the greater the value of the toxicity-economic impact factor term, which can highlight the high risk. By introducing the residual limit standard, the misjudgment of low-toxicity pesticides can be inhibited.

[0108] S14, obtaining an emergency index according to the network heat factor term, the toxicity-economic impact factor term, and the seasonal risk coefficient.

[0109] Specifically, the emergency index is expressed as wherein α1, α2, and α3 represent weight coefficients of the network heat factor term, the toxicity-economic impact factor term, and the seasonal risk coefficient, respectively, S i represents the seasonal risk coefficient of the pesticide i to be detected.

[0110] When a public opinion related to the pesticide i to be detected occurs, the weight coefficient of the weight network heat factor term can be automatically increased (generally, the default range is 0.3-0.5) to quickly respond to public concerns, for example, α1 can be increased from the original 0.5 to a higher 0.7 or 0.8. The weight coefficient α2 of the toxicity-economic impact factor term is set to 0.4-0.6 by default. For the detection service of exported agricultural products, the value of α2 can be appropriately increased to strengthen the compliance review.

[0111] Q-learning or DQN algorithm can be used, and α1, α2, and α3 can be continuously adjusted and optimized according to historical detection data and actual effects, so that the system can adapt to different scene requirements: 1. For the regular pesticide detection scene, the weight coefficients α1, α2, and α3 are evenly distributed; 2. For the environment of sudden public opinion, the weight coefficient of the network heat factor term is increased; 3. For the seasonal high-risk scene, the weight coefficient of the seasonal risk coefficient is increased.

[0112] As a model-free reinforcement learning algorithm, Q-learning stores the values of state-action pairs by constructing a Q table, which can preliminarily optimize the weight coefficients α1, α2, and α3. When the state space of the system becomes complex, traditional Q-learning is difficult to cope with. The DQN (Deep Q-Network) algorithm approximates the Q function through a deep neural network, which performs well in handling complex scenes in the pesticide detection system, can consider the influence of multiple factors on the weight coefficients at the same time, and is more suitable for handling high-dimensional state space.

[0113] For the obtained emergency index E i , normalization processing can be performed. As shown in Figure 9 , the emergency indexes of different pesticides to be detected are often different. It should be noted that Figure 9The figure only shows one of the different comparison diagrams of the emergency index of the pesticide to be detected. Due to the changes of the network heat factor, the toxicity-economic impact factor and the seasonal risk coefficient and the corresponding weight coefficient, the emergency index of the pesticide to be detected also changes synchronously.

[0114] The function formula By quantifying the emergency of pesticide detection in multiple dimensions, the problem of "important task response lag" in traditional detection can be solved. The heterogeneous data such as public opinion risk, toxicity hazard and economic impact are fused into a single index, which can guide intelligent allocation of detection resources, realize dynamic priority sorting, provide mathematical basis for collaborative scheduling of chromatograph cluster, ensure that high-urgent tasks occupy high-availability devices first, and can be used as a basis for resource optimization. By dynamically adjusting the weight coefficients of the seasonal risk coefficient and the network heat factor, potential public safety events can be responded to in advance, and the risk warning capability is possessed.

[0115] S2, real-time monitoring of the state parameters of each chromatograph, and obtaining the comprehensive availability of the chromatograph according to the state parameters. The state parameters include but are not limited to continuous running time, remaining life and detection queue length.

[0116] As a preferred technical solution, as shown in the figure, Figure 3 In step S2, the specific method for obtaining the comprehensive availability of the chromatograph includes:

[0117] S21, obtaining the continuous running time, and obtaining the running time item for reflecting the fatigue degree of the chromatograph according to the continuous running time.

[0118] Suppose t j , t max respectively represent the continuous running time of the chromatograph j and the maximum safe running time limit, then the running time item for reflecting the fatigue degree of the chromatograph can be represented as The continuous running time unit is hour or day, which is used for device fatigue to avoid performance degradation or failure risk caused by long-time running. The maximum safe running time limit is generally defined by the manufacturer or experimental specification.

[0119] Suppose the continuous running time and the maximum safe running time of the chromatograph j are 40 hours and 100 hours respectively, then The smaller the value of the running time item, the more fatigued the corresponding chromatograph is, and accordingly, its comprehensive availability will be lower.

[0120] S22, obtaining the remaining life and the theoretical life of the current chromatographic column, and obtaining the remaining life item for reflecting the aging degree of the chromatographic column according to the remaining life and the theoretical life.

[0121] Suppose A j , A newrespectively, the remaining life term for reflecting the aging degree of the chromatographic column can be expressed as The theoretical life can be understood as the theoretical life value of a new chromatographic column, which can be set to 1 by standardization processing. Correspondingly, the remaining life of the current chromatographic column of chromatograph j is processed by standardization, and its range is set to 0-1. Specifically, the remaining life can be obtained by the number of times of using the chromatographic column and / or the degree of pollution.

[0122] S23, obtain the length of the to-be-inspected queue, and obtain the queue pressure index term for reflecting the pressure of the chromatograph task queue according to the length of the to-be-inspected queue.

[0123] Suppose d j The length of the to-be-inspected queue of chromatograph j is represented by e, the natural constant is represented by λ, and the queue pressure index term for reflecting the pressure of the chromatograph task queue can be expressed as The length of the to-be-inspected queue of chromatograph j can be understood as the number of tasks, and the longer the queue, the more significant the exponential decay, and λ can be set according to experience.

[0124] S24, obtain the comprehensive availability according to the running time term, the remaining life term and the queue pressure index term.

[0125] Specifically, the comprehensive availability is expressed as Wherein, w1, w2, w3 represent the weight coefficients of the running time term, the remaining life term and the queue pressure index term, respectively.

[0126] w1, w2, w3 can be dynamically adjusted by reinforcement learning to balance the timeliness of detection and the service life of the equipment, for example, for urgent detection tasks, w1 and w3 can be increased to prioritize scheduling low-load chromatographs; for tasks that need long-term monitoring, w2 is increased to prolong the service life of the chromatographic column.

[0127] As shown in Figure 10 Due to the differences in the running time term, the remaining life term and the queue pressure index term, different chromatographs often have different comprehensive availabilities. In order to facilitate calculation, the comprehensive availability can be normalized.

[0128] In general, the comprehensive availability function is used to evaluate the comprehensive availability of the chromatograph in real time, by quantifying the three core parameters of the running time, the service life of the chromatographic column and the task queue pressure, a 0-1 normalized value is dynamically generated, which can balance the performance, life maintenance and task response efficiency of the equipment. It integrates the physical state (life / time) of the equipment and the task load, breaks through the traditional scheduling mode based only on the queue length, and can guide the intelligent allocation of pesticide detection tasks.

[0129] S3, obtaining the objective function according to the emergency index and the comprehensive availability.

[0130] As a preferred technical solution, as shown in step S3, the specific method for obtaining the objective function includes: Figure 4

[0131] S31, obtaining an efficiency item for representing the matching benefit between the pesticide detection task and the chromatograph according to the emergency index and the comprehensive availability.

[0132] The efficiency item is expressed as which can be used to maximize the matching degree of high-emergency tasks and high-availability devices. Wherein, E i represents the emergency index of the pesticide i to be detected, Q j represents the comprehensive availability of the chromatograph j, x ij is a binary decision variable and x ij = 1 indicates that the task i is assigned to the chromatograph j.

[0133] S32, obtaining a preset load balancing penalty factor and an ideal average queue length, and obtaining a load penalty item according to the load balancing penalty factor, the ideal average queue length, the total number of available chromatographs and the total number of pesticides to be detected.

[0134] The load penalty item is expressed as which is used to constrain the deviation of the number of tasks assigned to a single chromatograph from the ideal average queue length to avoid uneven resource allocation. Wherein, μ represents the load balancing penalty factor, N and M respectively represent the total number of available chromatographs and the total number of pesticides to be detected. μ>0, the greater the value, the more stringent the balance requirement. represents the total number of tasks assigned to the chromatograph j, and the ideal average queue length can be set according to experience or actual situation, or can be calculated according to the formula .

[0135] S33, obtaining the objective function according to the efficiency item and the load penalty item.

[0136] Specifically, the objective function is expressed as For a sudden high-emergency task, the emergency index increases sharply, the efficiency item weight increases, and the chromatograph resource with high comprehensive availability is preferentially occupied to shorten the key task detection time. When is greater than the ideal average queue length , the load penalty item increases, so the corresponding chromatograph can be avoided to avoid the risk of device overload and reduce the life loss of the chromatographic column. When is less than the ideal average queue length , the load penalty item can promote other pesticide detection tasks to be assigned to the corresponding chromatograph, thereby improving the overall device utilization efficiency of the cluster chromatograph.​

[0137] Preferably, when the objective function can be rewritten as to reduce the complexity of solving and further improve the response speed of the system. Wherein, δ j represents the preset queue length threshold of the chromatograph j.

[0138] The efficiency term aims to promote high-urgency tasks (such as sudden public opinion pesticides) to be preferentially matched with high-availability devices (such as newly replaced chromatographic columns). The load penalty term aims to suppress the phenomenon of "resource contention" and prevent multiple tasks from occupying a small number of high-performance devices, thereby prolonging the service life of low-end devices.

[0139] When the comprehensive availability of a chromatograph j is less than a preset availability threshold, such as 0.2, or the continuous running time is greater than the theoretical service life, the task allocation of the chromatograph j can be reduced until it enters the maintenance queue. For the urgency index and the comprehensive availability, they can be uploaded based on blockchain technology, and the global resource pooling scheduling can be realized through on-chain shared data updates.

[0140] When the remaining service life A j of the current chromatographic column of the chromatograph j is less than a preset service life threshold, such as 0.2, the load balancing penalty factor can be increased to preferentially protect relatively fragile chromatographs; and when public opinion related to the pesticide to be detected explodes, the load balancing penalty factor can be appropriately reduced to allow temporary load tilt to speed up critical tasks.

[0141] In some cases, constraints can be set, such as task mutual exclusion constraints, so that some pesticides cannot be used on the same device to avoid cross contamination. If a pesticide detection task requires a specific detection method, the binary decision variable x ij of the chromatograph incompatible with the pesticide detection task is equal to 0.

[0142] The objective function aims to achieve the dual goals of prioritizing emergency tasks and efficiently utilizing device resources, that is, under the premise of meeting device constraints, dynamically assigning pesticide detection tasks (sorted by urgency) to the most suitable chromatograph (evaluated by availability) to maximize global detection efficiency. This not only prioritizes high-urgency tasks and high-availability devices to shorten critical detection periods, but also avoids overloading or idling chromatographs through the load penalty term to prolong device service life.

[0143] S4, according to the objective function, realizing cluster collaborative control of the chromatographs.

[0144] Specifically, based on the objective function, after maximizing the global detection efficiency, the pesticide to be detected is assigned to the corresponding chromatograph for detection, thereby realizing cluster collaborative control of the chromatographs.

[0145] Exemplarily, assuming N=2, M=3, i.e., the total number of available chromatographs is 2, the total number of pesticides to be detected is 3, the emergency indexes E1, E2, E3 of the 3 pesticides to be detected are 0.8, 0.6, and 0.9 respectively, the comprehensive availabilities Q1, Q2 of the two chromatographs are 0.7 and 0.5 respectively, the load balancing penalty factor is set to 0.3, and the ideal average queue length is 2.

[0146] Two distribution schemes are considered, and distribution scheme one is shown in the following table:

[0147] Pesticide / Chromatograph Chromatograph 1 Chromatograph 2 Pesticide 1 1 0 Pesticide 2 0 1 Pesticide 3 1 0

[0148] The objective function is specifically expressed as The efficiency term The queue length of chromatograph 1 is 2, The queue length of chromatograph 2 is 1,

[0149] The load penalty term The final objective function value = 1.49-0.3 = 1.19.

[0150] Distribution scheme two is shown in the following table:

[0151] Pesticide / Chromatograph Chromatograph 1 Chromatograph 2 Pesticide 1 1 0 Pesticide 2 0 1 Pesticide 3 0 1

[0152] The objective function is specifically expressed as The efficiency term The queue length of chromatograph 1 is 1, The queue length of chromatograph 2 is 2,

[0153] The load penalty term The final objective function value = 1.31-0.3 = 1.01.

[0154] Since the objective function value of distribution scheme one is greater than that of distribution scheme two, distribution scheme one is better, which balances the high emergency task distribution and load balancing.

[0155] Exemplarily, assuming N=2, M=3, i.e., the total number of available chromatographs is 2, the total number of pesticides to be detected is 3, the emergency indexes E1, E2, E3 of the 3 pesticides to be detected are 0.8, 0.6, and 0.9 respectively, the comprehensive availabilities Q1, Q2 of the two chromatographs are 0.7 and 0.9 respectively, the load balancing penalty factor is set to 0.5, and the ideal average queue length is 2.

[0156] Suppose the distribution scheme is shown in the following table:

[0157] Pesticide / Chromatograph Chromatograph 1 Chromatograph 2 Pesticide 1 1 0 Pesticide 2 0 1 Pesticide 3 1 0

[0158] The objective function is specifically expressed as Efficiency term The queue length of the chromatograph 1 is 2, The queue length of the chromatograph 2 is 1,

[0159] Load penalty term The final objective function value = 1.73-0.5 = 1.23.

[0160] Overall, the allocation scheme allocates the high-urgency task (pesticide 3) to the high-availability chromatograph (chromatograph 2), which balances between high-urgency tasks and balanced load, and the total benefit is relatively high.

[0161] Suppose the allocation scheme is as shown in the following table:

[0162] Pesticide / Chromatograph Chromatograph 1 Chromatograph 2 Pesticide 1 1 0 Pesticide 2 1 0 Pesticide 3 1 0

[0163] The efficiency term The queue length of the chromatograph 1 is 3, The queue length of the chromatograph 2 is 0, The load penalty term The final objective function value = 1.61-1.5 = 0.11.

[0164] Although the benefit term score is high, the value of the load penalty term is large due to the overall load imbalance, and the total benefit is greatly reduced, indicating that the load balancing penalty factor can effectively suppress resource waste. The lower the load balancing penalty factor, the more the system can tolerate the imbalance of load allocation.

[0165] In summary, the chromatograph cluster cooperative control method of the pesticide detection service obtains the urgency index and the comprehensive availability, obtains the target function according to the urgency index and the comprehensive availability, and finally realizes the cluster cooperative control of the chromatograph based on the target function. It can dynamically allocate pesticide detection tasks (sorted according to the urgency index) to the most suitable chromatograph (i.e. according to the comprehensive availability), maximize the global detection efficiency, break through the traditional chromatograph cluster scheduling detection strategy relying on "first-come-first-detection" or fixed cycle, and quickly respond to sudden pesticide safety events.

[0166] As Figure 5 shown, the present application also provides a chromatograph cluster cooperative control system for pesticide detection service for realizing the chromatograph cluster cooperative control method of the pesticide detection service, which comprises an urgency index acquisition module, a comprehensive availability acquisition module, a target function acquisition module and a cluster cooperative control module.

[0167] The urgency index acquisition module is used to acquire the urgency index of the pesticide to be tested. As a preferred technical solution, the urgency index acquisition module includes a first acquisition unit, a second acquisition unit, a third acquisition unit, and a fourth acquisition unit.

[0168] The first acquisition unit is used to acquire the online popularity, intrinsic characteristics, and seasonal risk coefficient of the pesticide to be tested; the second acquisition unit is used to acquire the online popularity factor based on the online popularity of the pesticide to be tested. The third acquisition unit is used to acquire the toxicity-economic impact factor based on the toxicity level, crop economic value coefficient, and residue limit standard; the fourth acquisition unit is used to acquire the urgency index based on the online popularity factor, the toxicity-economic impact factor, and the seasonal risk coefficient; wherein, the intrinsic characteristics include the toxicity level of the pesticide to be tested, the corresponding crop economic value coefficient, and the residue limit standard.

[0169] The overall availability acquisition module is used to monitor the status parameters of each chromatograph in real time and obtain the overall availability of the chromatograph based on the status parameters; the objective function acquisition module is used to obtain the objective function based on the urgency index and the overall availability.

[0170] As a preferred technical solution, the objective function acquisition module includes a benefit term acquisition unit, a penalty term acquisition unit, and an objective function acquisition unit.

[0171] The benefit term acquisition unit is used to obtain an efficiency term representing the matching benefit between pesticide detection tasks and chromatographs based on the urgency index and overall availability. The penalty term acquisition unit is used to obtain a preset load balancing penalty factor and ideal average queue length, and obtain a load penalty term based on the load balancing penalty factor, ideal average queue length, total number of available chromatographs, and total number of pesticides to be detected. The objective function acquisition unit is used to obtain an objective function based on the efficiency term and the load penalty term. The cluster collaborative control module is used to implement cluster collaborative control of the chromatographs based on the objective function.

[0172] In summary, the chromatograph cluster collaborative control system can dynamically allocate pesticide detection tasks (ranked by urgency index) to the most suitable chromatograph (i.e., based on overall availability), thereby maximizing global detection efficiency. This breaks through the traditional chromatograph cluster scheduling strategy that relies on "first-come, first-served" or fixed-cycle detection, and can quickly respond to sudden pesticide safety incidents.

[0173] In one embodiment, the present invention is experimentally simulated using the following key parameters.

[0174] Experimental Design and Parameter Configuration

[0175] 1. Hardware configuration: A detection cluster consisting of 20 chromatographs. For example... Figure 6 The diagram shows the continuous operating time of 20 chromatographs.Figure 7 Figure 4 shows a schematic diagram of the life status of the chromatographic columns of 20 chromatographs, Figure 8 Figure 5 shows a schematic diagram of the queue length of 20 chromatographs to be detected.

[0176] 2. Task load: simulate 0-50 pesticide residue detection tasks per day, task arrival obeys Poisson distribution, average interval time is 15 minutes.

[0177] 3. Pesticide sample characteristics:

[0178] 3.1 Toxicity level distribution: level 1 (10%), level 2 (25%), level 3 (40%), level 4 (20%), level 5 (5%).

[0179] 3.2 Residue threshold range: 0.01-1 mg / kg, lognormal distribution.

[0180] 3.3 Crop relevance: rice (30%), wheat (25%), vegetables (35%), others (10%).

[0181] 4. Weight coefficient configuration: initial weight coefficients a1, a2, a3 in the conventional monitoring mode are 0.15, 0.7 and 0.15, respectively.

[0182] 5. Experimental period: simulate continuous system operation for 7 days, 8 hours of working time per day.

[0183] Experimental process and data collection

[0184] During the experiment, the system operates according to the following steps:

[0185] 1. Task generation: randomly generate pesticide detection tasks according to the configured parameters, including pesticide type, toxicity level, residue limit standard and other attributes

[0186] 2. Emergency index calculation: the system considers pesticide characteristics, network popularity and seasonal factors to calculate the emergency index of each detection task in real time.

[0187] 3. Equipment state monitoring: real-time collection of continuous running time of chromatograph, remaining life of chromatographic column and queue length to be detected, calculation of comprehensive availability.

[0188] 4. Task allocation decision: based on the emergency index and the comprehensive availability of the equipment, the system generates the optimal task allocation scheme based on the objective function.

[0189] 5. Data recording: the system records the task completion time, equipment utilization, load balancing degree and other key performance indicators.

[0190] Experimental results and analysis

[0191] The experimental results show that the dynamic priority model (i.e., the urgency index function, which dynamically adjusts and optimizes the weight coefficients α1, α2, α3 based on historical detection data and actual effects) performs well under normal load, as shown in the following table:

[0192] Performance Index Traditional Static Scheduling Dynamic Priority Model Improvement Ratio Average Task Completion Time 5.8 hours 3.7 hours 36.2％ Device Utilization 72％ 92％ 27.8％ Load Balancing Degree 68％ 91％ 33.8％ High Priority Task Response Time 4.2 hours 2.1 hours 50.0％

[0193] The data shows that the dynamic priority model can significantly improve the overall performance of the system, especially in the response time of high-priority tasks, which is improved by 50%.

[0194] On the third day of the experiment, the system detected a slight increase in the heat of glyphosate-related network public opinion, and automatically adjusted the network heat weight of the pesticide from 0.15 to 0.25. The urgency index was correspondingly increased, and the detection response time was shortened from 4.5 hours to 2.8 hours.

[0195] This case demonstrates the system's ability to sensitively capture small fluctuations in network public opinion in normal scenarios, reflecting the adaptive characteristics of the dynamic priority model.

[0196] Resource scheduling efficiency analysis

[0197] In the experiment, it was observed that the resource scheduling of the chromatograph cluster showed high efficiency in load balancing:

[0198] 1. Task allocation balance: The standard deviation of the task queue length of the 20 devices is only 2.3, indicating that the system can effectively avoid device overload or idling.

[0199] 2. Device state optimization: The system intelligently allocates detection tasks based on the remaining life of the chromatographic column and the continuous running time, prolonging the service life of the device and reducing maintenance costs.

[0200] 3. Pesticide property matching: Different types of pesticides are allocated to the most suitable chromatograph, improving detection accuracy and efficiency.

[0201] The simulation experiment in the normal scenario proves that the pesticide detection system based on the dynamic priority model can achieve efficient resource scheduling and task processing under normal load, providing reliable technical support for the supervision of agricultural product quality and safety.

[0202] In addition, the invention also conducts a simulation experiment in a sudden public opinion scenario: In the sudden public opinion scenario of pesticide safety incidents, the pesticide residue detection system faces challenges such as a surge in detection demand, high public attention, and strict response time requirements. This embodiment also simulates a glyphosate pesticide safety incident to verify the system's emergency response capability in a public opinion crisis.

[0203] Experimental design and parameter configuration

[0204] 1. Public opinion event simulation: simulate a public opinion event of glyphosate pesticide exceeding the standard in vegetables in a certain area, the network heat rises sharply from 0.15 to 0.85

[0205] 2. System configuration: 20 chromatographs constitute a detection cluster; the weight coefficient is automatically adjusted to the sudden public opinion mode: α1, α2, α3 are set to 0.35, 0.6 and 0.05 respectively.

[0206] 3. Task load: 50 regular detection tasks per day; 30 new ones within 24 hours after the burst

[0207] Glyphosate-related emergency detection tasks.

[0208] System response process

[0209] 1. Public opinion monitoring and early warning: the system detects that the discussion volume related to glyphosate has increased by 300% within 2 hours through monitoring network public opinion, and automatically triggers the public opinion early warning mechanism.

[0210] 2. Dynamic weight adjustment: the system identifies it as a sudden public opinion scene, and increases the weight coefficient of the network heat factor from 0.15 to 0.35, greatly increasing the emergency index of glyphosate-related detection tasks.

[0211] 3. Preemptive resource scheduling: the system automatically suspends 8 low-urgency tasks, allocates 4 high-performance liquid chromatographs exclusively to glyphosate detection tasks, and starts a parallel processing mechanism to maximize detection efficiency.

[0212] 4. Chromatograph cluster collaborative control, achieving dual optimization of load balancing and detection efficiency.

[0213] Experimental results and analysis

[0214] Performance Index Traditional Static Scheduling Dynamic Priority Model Improvement Ratio Glyphosate Detection Response Time 24 hours 4 hours 83.3％ First Batch Result Release Time 36 hours 6 hours 83.3％ Completion Rate within 24 hours 40％ 95％ 137.5％ System Resource Utilization 65％ 96％ 47.7％

[0215] The data shows that in the sudden public opinion scenario, the dynamic priority model performs well, especially in the key response time indicators, with the first batch of detection results published in 6 hours instead of the traditional 36 hours, providing strong support for responding to public concerns in a timely manner.

[0216] In summary, the simulation experiment of the sudden public opinion scenario proves that the system can effectively deal with pesticide safety incidents, provide timely and accurate detection data for government departments, help control public opinion crises, and protect public health and social stability.

[0217] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, several modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for collaborative control of a chromatograph cluster for pesticide detection services, characterized in that, The method comprises the following steps: An emergency index of the pesticide to be detected is obtained; Real-time monitoring is performed on state parameters of each chromatograph, and a comprehensive availability of the chromatograph is obtained according to the state parameters; A target function is obtained according to the emergency index and the comprehensive availability; Clustered collaborative control of the chromatographs is realized according to the target function; The specific method for obtaining the target function comprises: An efficiency item representing matching benefits between the pesticide detection task and the chromatograph is obtained according to the emergency index and the comprehensive availability; A preset load balancing penalty factor and an ideal average queue length are obtained, and a load penalty item is obtained according to the load balancing penalty factor, the ideal average queue length, a total number of available chromatographs and a total number of pesticides to be detected; The target function is obtained according to the efficiency item and the load penalty item. The specific method for obtaining the emergency index of the pesticide to be detected comprises: A network popularity of the pesticide to be detected, self characteristics and a seasonal risk coefficient are obtained; A network popularity factor item is obtained according to the network popularity of the pesticide to be detected; A toxicity-economic influence factor item is obtained according to a toxicity grade, a crop economic value coefficient and a residue limit standard; The emergency index is obtained according to the network popularity factor item, the toxicity-economic influence factor item and the seasonal risk coefficient. The self characteristics comprise the toxicity grade, the crop economic value coefficient and the residue limit standard of the pesticide to be detected.

2. The method of claim 1, wherein the method is characterized by, The specific method for obtaining the comprehensive availability of the chromatograph comprises: A continuous running duration is obtained, and a running duration item reflecting fatigue of the chromatograph is obtained according to the continuous running duration; A remaining life of a current chromatographic column and a theoretical life are obtained, and a remaining life item reflecting an aging degree of the chromatographic column is obtained according to the remaining life and the theoretical life; A to-be-detected queue length is obtained, and a queue pressure index item reflecting task queue pressure of the chromatograph is obtained according to the to-be-detected queue length; The comprehensive availability is obtained according to the running duration item, the remaining life item and the queue pressure index item. The state parameters comprise the continuous running duration, the remaining life and the to-be-detected queue length.

3. The method of claim 2, wherein the method further comprises: The objective function is expressed as ; in, These represent the efficiency term and the load penalty term, respectively. Indicates the pesticide to be tested The urgency index, Indicates chromatograph Overall availability For binary decision variables and when Indicates the task Dispensed to chromatograph , This represents the load balancing penalty factor. This represents the ideal average queue length. These represent the total number of available chromatographs and the total number of pesticides to be detected, respectively.

4. The method of claim 3, wherein the method further comprises: The urgency index is expressed as ; in, These represent the online popularity factor and the toxicity-economic impact factor, respectively. These represent the weighting coefficients of the online popularity factor, the toxicity-economic impact factor, and the seasonality risk coefficient, respectively. Indicates the pesticide to be tested The online popularity Indicates the preset pesticide to be detected The baseline value for heat, Indicates the pesticide to be tested Toxicity level, Indicates the pesticide to be tested The crop economic value coefficient, Indicates the pesticide to be tested The residue limit standard, Indicates the pesticide to be tested The seasonal risk coefficient.

5. The method of claim 4, wherein the method further comprises: The overall availability is expressed as ; wherein, respectively represent a runtime item, a remaining lifetime item, and a queue pressure index item, respectively represent weight coefficients of the runtime item, the remaining lifetime item, and the queue pressure index item, respectively represent a continuous runtime of the chromatograph and a preset maximum safe runtime limit, respectively represent a current column of the chromatograph and a theoretical lifetime, represent a queue length of the chromatograph to be inspected, represent a natural constant, represent a queue pressure decay coefficient.

6. A chromatograph cluster cooperative control system for a pesticide detection service, for implementing the pesticide detection service chromatograph cluster cooperative control method according to any one of claims 1 to 5, characterized by, The method comprises: An emergency index obtaining module is configured to obtain an emergency index of a pesticide to be detected; A comprehensive availability obtaining module is configured to perform real-time monitoring on state parameters of each chromatograph, and obtain a comprehensive availability of the chromatograph according to the state parameters; A target function obtaining module is configured to obtain a target function according to the emergency index and the comprehensive availability; A clustered collaborative control module is configured to realize clustered collaborative control of the chromatographs according to the target function.

7. The collaborative control system of a cluster of chromatographs for a pesticide detection service of claim 6, wherein, The target function obtaining module comprises: An efficiency item obtaining unit is configured to obtain an efficiency item representing matching benefits between a pesticide detection task and a chromatograph according to an emergency index and a comprehensive availability; A penalty item obtaining unit is configured to obtain a preset load balancing penalty factor and an ideal average queue length, and obtain a load penalty item according to the load balancing penalty factor, the ideal average queue length, a total number of available chromatographs and a total number of pesticides to be detected; A target function obtaining unit is configured to obtain a target function according to the efficiency item and the load penalty item.

8. The collaborative control system of a cluster of chromatographs for a pesticide detection service of claim 6, wherein, The emergency index obtaining module comprises: A first obtaining unit is configured to obtain a network popularity of a pesticide to be detected, self characteristics and a seasonal risk coefficient. The second obtaining unit is configured to obtain a network heat factor term according to network heat of the pesticide to be detected; The third obtaining unit is configured to obtain a toxicity-economic impact factor term according to the toxicity grade, the crop economic value coefficient and the residue limit standard; The fourth obtaining unit is configured to obtain an emergency degree index according to the network heat factor term, the toxicity-economic impact factor term and the seasonal risk coefficient. The self characteristics include the toxicity grade, the corresponding crop economic value coefficient and the residue limit standard of the pesticide to be detected.

Images