An intelligent multi-state sensing ventilation control system and method based on simulation

Abstract

The invention discloses an intelligent multi-state perception ventilation control system and method based on simulation, which relates to the technical field of ventilation control, including: obtaining historical data of a fermentation stack, and dividing the historical data of the fermentation stack into historical data when there is no leakage and historical data when there is leakage according to whether there is leakage in an air supply pipeline; training a parameter prediction model of the fermentation stack based on the historical data when there is no leakage; verifying the parameters of the fermentation stack to determine whether there is a leakage event in the air supply pipeline; improving the parameter simulation accuracy of the fermentation stack by considering the hysteresis of the air supply operation and the influence of residual gas; determining whether a leakage event occurs according to the error between the simulated value and the measured value of the fermentation stack parameter, providing support for leakage detection, and improving the efficiency of leakage detection; screening the leakage position to further improve the detection efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of ventilation control, and in particular to an intelligent multi-state perception ventilation control system and method based on simulation. Background Art

[0002] NCS intelligent molecular membrane is a technology used for resource utilization of livestock and poultry manure. It converts livestock and poultry manure into organic fertilizer or cattle bedding through film-covered composting fermentation. The NCS intelligent molecular membrane system is mainly composed of an air supply system, a monitoring system and an NCS intelligent molecular membrane. It is mainly used in agriculture and resource utilization of livestock and poultry manure. The micro-pressure air supply system plays an important role in the aerobic fermentation process of organic waste. The leakage of air supply ducts is a problem that needs to be paid attention to in the micro-pressure air supply system. It seriously affects the efficiency of the entire system and the quality of aerobic fermentation. The existing technology cannot detect leakage of ventilation ducts all the time, and there is a problem of untimely leakage of air supply ducts. Therefore, how to improve the detection efficiency of air supply ducts has become an urgent problem to be solved. Summary of the invention

[0003] The purpose of the present invention is to provide an intelligent multi-state perception ventilation control system and method based on simulation to solve the problems raised in the above background technology.

[0004] In one aspect of the present invention, there is provided an intelligent multi-state sensing ventilation control method based on simulation, comprising:

[0005] S11, obtaining historical data of the fermentation stack, and according to whether there is a leak in the air supply pipeline, the historical data of the fermentation stack is divided into historical data when there is no leak and historical data when there is a leak;

[0006] S12, training a parameter prediction model of the fermentation pile based on historical data when there is no leakage;

[0007] S13, based on the parameter prediction model of the fermentation stack, verify the parameters of the fermentation stack to determine whether there is a leakage event in the air supply duct; if there is a leakage event, perform a leakage detection on the air supply duct; if there is no leakage event, continue to verify the parameters of the fermentation stack.

[0008] In step S12, the training parameter prediction model of the fermentation pile body further includes the following steps:

[0009] S21, based on historical data when there is no leakage, determining the time T when the air supply operation affects the fermentation stack;

[0010] S22, the historical data without leakage is divided into a training set and a test set. In the training set, all fermentation pile parameters, air supply parameters g, organic waste type type and fermentation time at time t and after time t are taken as input. As output, the For the i-th fermentation pile parameter, train the prediction model of the i-th fermentation pile parameter, and use the historical data of the test set to verify the prediction model of the i-th fermentation pile parameter; obtain the prediction model of all fermentation pile parameters; if air supply is being carried out at time t, the air supply parameter g is 1, if air supply is not being carried out at time t, the air supply parameter g is 0; distinguish the types of organic waste by assigning different discrete values ​​to type.

[0011] Starting air supply does not affect the fermentation pile immediately, but requires a certain amount of time; similarly, after stopping air supply, the residual gas in the air supply duct will not disappear immediately, but will continue to affect the fermentation pile for a period of time.

[0012] Organic waste fermentation is usually divided into four stages: heating stage, high-temperature fermentation stage, medium-temperature fermentation stage and cooling stage. The parameter models of different stages are different. Therefore, the time of the fermentation process needs to be considered to predict the fermentation pile parameters. At the same time, different types of organic waste have different fermentation cycles. For example, the fermentation cycle of cow dung is 7-15 days, and the fermentation cycle of pig manure is about 25 days, which needs to be distinguished.

[0013] In step S21, the step of determining the time T that the air supply operation affects the fermentation stack further includes the following steps:

[0014] Based on the historical data when there is no leakage, the dividing point c1 between air supply operation and no air supply operation and the dividing point c2 between no air supply operation and air supply operation are determined, and the parameter data between the dividing points c1 and c2 are used as the first input feature, and the parameter data between the dividing points c2 and c1 are used as the second input feature; the air supply operation and the no air supply operation are switched alternately, and there are several groups of first input features and second input features;

[0015] For each group of first input features, supervised classification is used for classification. In a group of first input features, the 1st, 2nd, 3rd, ...T1+1th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The first classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T1+1, and the classification accuracy A1 of the negative examples is calculated by the following formula: , N21 is the number of negative examples in the group, N11 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT1 of T1 obtained according to the first input feature of all groups;

[0016] For each group of second input features, supervised classification is used for classification. In a group of second input features, the 1st, 2nd, 3rd, ...T2th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The second classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T2+1, and the classification accuracy A2 of the negative examples is calculated by the following formula: , N22 is the number of negative examples in the group, N12 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT2 of T2 obtained according to the second input feature of all groups, round up the means of AT1 and AT2, and get the time T that the air supply operation affects the fermentation pile.

[0017] After the air supply is stopped, the impact on the fermentation pile is continuous. The fermentation pile is affected by the residual gas in the first period, and the residual gas is completely consumed in the second period. Similarly, after the air supply is started, the impact on the fermentation pile is continuous. Binary classification is used to find the impact time T, and positive and negative examples are required to be continuously distributed.

[0018] In step S13, the process of determining whether there is a leakage in the air supply duct further comprises the following steps:

[0019] S41, in the historical data when there is no leakage, the parameter data of T consecutive fermentation stacks, the air supply parameter g, the type of organic waste type and the fermentation time are input as input to the prediction model of the i-th fermentation stack parameter, and the i-th fermentation stack parameter simulation value is obtained, and the error between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter is calculated; there are K1 time windows in the historical data, and the error between K1 measured values ​​and the simulated values ​​is obtained;

[0020] In the historical data when there is leakage, the parameter data of T consecutive fermentation piles, the air supply parameter g, the type of organic waste type and the fermentation time are used as inputs and input into the prediction model of the i-th fermentation pile parameter to obtain the i-th fermentation pile parameter simulation value, and calculate the error between the measured value and the simulated value of the T+1-th i-th fermentation pile parameter; there are K2 time windows in the historical data, and the errors between K2 measured values ​​and simulated values ​​are obtained;

[0021] S42, obtaining all parameter data of the fermentation stack, inputting parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of the i-th fermentation stack parameter, obtaining the T+1-th i-th fermentation stack parameter simulation value, and calculating the error e between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter;

[0022] S43, calculate the leakage probability of the air supply duct , , is the leakage probability corresponding to the i-th fermentation pile parameter, , The number of historical data with or without leakage in which the error between the measured value and the simulated value is greater than or equal to e; is the number of historical data in which the error between the measured value and the simulated value is greater than or equal to e in the historical data when there is leakage; if the leakage probability of the air supply duct If the value is greater than or equal to the preset threshold, it is determined that there is a leakage event in the air supply duct; otherwise, it is determined that there is no leakage event in the air supply duct.

[0023] When no leakage occurs, there is a certain error between the simulation value and the measured value. After the leakage occurs, the model is no longer applicable, and the error between the simulation value and the measured value will become larger. The leakage probability of the air supply duct is judged based on the error.

[0024] In step S13, the process of determining whether there is a leakage in the air supply duct further comprises the following steps:

[0025] Obtain the error e between the measured value and the simulated value of the fermentation stack parameter; determine the leakage position of the air supply duct and the error between the measured value and the simulated value corresponding to the leakage position in the historical data when there is a leakage; determine the wind resistance distribution of the air supply duct according to the installation situation and material properties of the air supply duct, and the wind resistance distribution is a fixed value; perform unsupervised classification on the error e between the measured value and the simulated value of the fermentation stack parameter and the error between the measured value and the simulated value of the leakage position of the air supply duct in the historical data, determine the classification cluster Z to which the error e between the measured value and the simulated value of the fermentation stack parameter belongs, find the leakage position and the wind resistance at the leakage position corresponding to the error in the historical data in the classification cluster Z, determine the minimum value min1 and maximum value max1 of the leakage position from the starting point of the air supply duct, and the minimum value min2 and maximum value max2 of the wind resistance, and include the air supply duct whose distance from the starting point of the air supply duct is between min1 and max1 and whose wind resistance is between min2 and max2 into the key detection range.

[0026] In the air supply duct, the pressure usually decreases, and the distance from the starting point of the air supply duct can reflect the pressure change; after the air supply duct is installed, the wind resistance value is fixed. Under different wind resistance values ​​and pressure conditions, the impact of leakage on the fermentation pile is different, that is, the parameter measurement values ​​are different, which is the error between the measured value and the simulated value after conversion; for this reason, the leakage location can be simply screened according to the error to speed up the detection speed.

[0027] In step S13, the air supply duct is leak-checked, which also includes the following steps:

[0028] First, the air supply ducts within the key inspection range are tested for leaks. If no leaks are detected, the air supply ducts at other locations are tested. If leaks are detected, the leaking locations are repaired.

[0029] In another aspect of the present invention, an intelligent multi-state perception ventilation control system based on simulation is provided, comprising a data monitoring module, a state analysis module, a data storage module, a detection module and a maintenance management module; the output end of the data monitoring module is connected to the input end of the data storage module, for obtaining parameter data of the fermentation stack and storing it in the data storage module; the output end of the state analysis module is connected to the input end of the detection module, and the integrity of the air supply duct is confirmed according to the parameter data of the fermentation stack, the fermentation time and the air supply condition; the output end of the data storage module is connected to the input end of the state analysis module, for storing the parameter data and air supply condition data during the fermentation process; the output end of the detection module is connected to the input end of the maintenance management module, for performing leakage detection on the air supply duct; the maintenance management module, when the detection module determines that there is a leakage in the air supply duct, contacts the management personnel to repair the air supply duct.

[0030] The data monitoring module also includes a sensor unit, a timer unit and a measurement and transmission unit; the sensor unit is used to obtain parameter data of the fermentation pile; the timer unit is used to determine the duration of the fermentation process; the measurement and transmission unit is used to determine the air supply condition of the air supply duct.

[0031] The state analysis module also includes an impact time determination unit, a fermentation stack parameter prediction unit, a simulation detection unit, a data dimension reduction unit and a leakage analysis unit; the impact time determination unit is used to determine the time window from the start of the air supply operation to the actual impact on the fermentation stack; the fermentation stack parameter prediction unit is used to train the prediction model of each fermentation stack parameter; the simulation detection unit is used to obtain the fermentation stack parameters in the absence of leakage; the leakage analysis unit determines the probability of a leakage event in the air supply duct based on the error between the simulated value and the measured value of the fermentation stack parameter; the data dimension reduction unit is used to perform dimensionality reduction processing on the data within the time window.

[0032] The simulation detection unit inputs the parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of each fermentation stack parameter, obtains the simulated value of the T+1th fermentation stack parameter when there is no leakage, and compares it with the actual value of the T+1th fermentation stack parameter to determine the error between the simulated value and the measured value of the fermentation stack parameter.

[0033] Compared with the prior art, the beneficial effects achieved by the present invention are: considering the hysteresis of the air supply operation and the influence of residual gas, the parameter simulation accuracy of the fermentation stack is improved; judging whether a leakage event occurs based on the error between the fermentation stack parameter simulation value and the measured value, providing support for leak detection and improving the efficiency of leak detection; screening the leak location to further improve the detection efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0035] Figure 1 It is a structural schematic diagram of an intelligent multi-state perception ventilation control system based on simulation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0036] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] In the embodiments of the present invention, please refer to Figure 1 , provides an intelligent multi-state perception ventilation control system based on simulation, including:

[0038] A data monitoring module, a state analysis module, a data storage module, a detection module and a maintenance management module; the output end of the data monitoring module is connected to the input end of the data storage module, and is used to obtain parameter data of the fermentation pile body and store it in the data storage module; the output end of the state analysis module is connected to the input end of the detection module, and the integrity of the air supply duct is confirmed according to the parameter data of the fermentation pile body, the fermentation time and the air supply condition; the output end of the data storage module is connected to the input end of the state analysis module, and is used to store the parameter data and air supply condition data during the fermentation process; the output end of the detection module is connected to the input end of the maintenance management module, and is used to perform leakage detection on the air supply duct; the maintenance management module, when the detection module determines that there is a leakage in the air supply duct, contacts the management personnel to repair the air supply duct.

[0039] The data monitoring module also includes a sensor unit, a timer unit and a measurement and transmission unit; the sensor unit is used to obtain parameter data of the fermentation stack; the timer unit is used to determine the duration of the fermentation process; the measurement and transmission unit is used to determine the air supply condition of the air supply duct. The state analysis module also includes an impact time determination unit, a fermentation stack parameter prediction unit, a simulation detection unit, a data dimension reduction unit and a leakage analysis unit; the impact time determination unit is used to determine the time window from the start of the air supply operation to the actual impact on the fermentation stack; the fermentation stack parameter prediction unit is used to train a prediction model for each fermentation stack parameter; the simulation detection unit is used to obtain the fermentation stack parameters in the absence of leakage; the leakage analysis unit determines the probability of a leakage event in the air supply duct based on the error between the simulated value and the measured value of the fermentation stack parameter; the data dimension reduction unit is used to perform dimension reduction processing on the data within the time window. The simulation detection unit inputs the parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of each fermentation stack parameter, obtains the simulated value of the T+1th fermentation stack parameter when there is no leakage, and compares it with the actual value of the T+1th fermentation stack parameter to determine the error between the simulated value and the measured value of the fermentation stack parameter.

[0040] In an embodiment of the present invention, an intelligent multi-state sensing ventilation control method based on simulation is provided, comprising:

[0041] S11, obtaining historical data of the fermentation stack, and according to whether there is a leak in the air supply duct, the historical data of the fermentation stack is divided into historical data when there is no leak and historical data when there is a leak.

[0042] S12, based on the historical data when there is no leakage, training the parameter prediction model of the fermentation pile; the fermentation pile parameters include temperature, humidity and oxygen concentration;

[0043] Training the parameter prediction model of the fermentation pile also includes the following steps:

[0044] First, determine the time T that the air supply operation affects the fermentation pile:

[0045] Based on the historical data when there is no leakage, the dividing point c1 between air supply operation and no air supply operation and the dividing point c2 between no air supply operation and air supply operation are determined, and the parameter data between the dividing points c1 and c2 are used as the first input feature, and the parameter data between the dividing points c2 and c1 are used as the second input feature; the air supply operation and the no air supply operation are switched alternately, and there are several groups of first input features and second input features;

[0046] For each group of first input features, supervised classification is used for classification. In a group of first input features, the 1st, 2nd, 3rd, ...T1+1th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The first classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T1+1, and the classification accuracy A1 of the negative examples is calculated by the following formula: , N21 is the number of negative examples in the group, N11 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT1 of T1 obtained according to the first input feature of all groups;

[0047] For each group of second input features, supervised classification is used for classification. In a group of second input features, the 1st, 2nd, 3rd, ...T2th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The second classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T2+1, and the classification accuracy A2 of the negative examples is calculated by the following formula: , N22 is the number of negative examples in the group, N12 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT2 of T2 obtained according to the second input feature of all groups, round up the means of AT1 and AT2, and get the time T that the air supply operation affects the fermentation pile.

[0048] For the 30 parameter data after the air supply is stopped, each parameter data includes temperature, humidity and oxygen concentration; first, the first parameter data is used as a positive example and the remaining 19 are used as negative examples to train a binary classification model. If the accuracy of the negative example meets the requirements, the first two parameter data are used as positive examples and the remaining 18 are used as negative examples to train the binary classification model until the accuracy of the negative example does not meet the requirements; the binary classification model can adopt models such as support vector machine, artificial neural network, and multi-layer perceptron.

[0049] Then train the model according to step S22:

[0050] S22, the historical data without leakage is divided into a training set and a test set. In the training set, all fermentation pile parameters, air supply parameters g, organic waste type type and fermentation time at time t and after time t are taken as input. As output, the For the i-th fermentation pile parameter, train the prediction model of the i-th fermentation pile parameter, and use the historical data of the test set to verify the prediction model of the i-th fermentation pile parameter; obtain the prediction model of all fermentation pile parameters; if air supply is being carried out at time t, the air supply parameter g is 1, if air supply is not being carried out at time t, the air supply parameter g is 0; distinguish the types of organic waste by assigning different discrete values ​​to type.

[0051] S13, based on the parameter prediction model of the fermentation stack, verify the parameters of the fermentation stack to determine whether there is a leakage event in the air supply duct; if there is a leakage event, perform a leakage detection on the air supply duct; if there is no leakage event, continue to verify the parameters of the fermentation stack.

[0052] When the time T when the air supply operation affects the fermentation pile is 10, in the training set, the fermentation pile parameter data, air supply parameter g, organic waste type type and fermentation time at moments 0-9 are input, and the temperature at the 10th moment is output. The time window is shifted, and the fermentation pile parameter data, air supply parameter g, organic waste type type and fermentation time at moments 1-10 are input, and the temperature at the 11th moment is output, and so on. The temperature prediction model can be trained using tools such as support vector machine regression, LSTM neural network, and CNN neural network; the humidity and oxygen content prediction models are trained in the same way as the temperature.

[0053] Determining whether there is a leakage in the air supply duct also includes the following steps:

[0054] S41, in the historical data when there is no leakage, the parameter data of T consecutive fermentation stacks, the air supply parameter g, the type of organic waste type and the fermentation time are input as input to the prediction model of the i-th fermentation stack parameter, and the i-th fermentation stack parameter simulation value is obtained, and the error between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter is calculated; there are K1 time windows in the historical data, and the error between K1 measured values ​​and the simulated values ​​is obtained;

[0055] In the historical data when there is leakage, the parameter data of T consecutive fermentation piles, the air supply parameter g, the type of organic waste type and the fermentation time are used as inputs and input into the prediction model of the i-th fermentation pile parameter to obtain the i-th fermentation pile parameter simulation value, and calculate the error between the measured value and the simulated value of the T+1-th i-th fermentation pile parameter; there are K2 time windows in the historical data, and the errors between K2 measured values ​​and simulated values ​​are obtained;

[0056] In the historical data when there is a leak, the output simulation value at time 10 is obtained based on the input at time 0-9, and the error is obtained by comparing it with the measured value at time 10; the output simulation value at time 11 is obtained based on the input at time 1-10, and the error is obtained by comparing it with the measured value at time 11, and so on to obtain the error data between the measured value and the simulation value;

[0057] S42, obtaining all parameter data of the fermentation stack, inputting parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of the i-th fermentation stack parameter, obtaining the T+1-th i-th fermentation stack parameter simulation value, and calculating the error e between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter;

[0058] S43, calculate the leakage probability of the air supply duct , , is the leakage probability corresponding to the i-th fermentation pile parameter, , The number of historical data with or without leakage in which the error between the measured value and the simulated value is greater than or equal to e; is the number of historical data in which the error between the measured value and the simulated value is greater than or equal to e in the historical data when there is leakage; if the leakage probability of the air supply duct If the value is greater than or equal to the preset threshold, it is determined that there is a leakage event in the air supply duct; otherwise, it is determined that there is no leakage event in the air supply duct.

[0059] In particular, determining whether there is a leakage event in the air supply duct also includes the following steps:

[0060] Obtain the error e between the measured value and the simulated value of the fermentation stack parameter; determine the leakage position of the air supply duct and the error between the measured value and the simulated value corresponding to the leakage position in the historical data when there is a leakage; determine the wind resistance distribution of the air supply duct according to the installation situation and material properties of the air supply duct, and the wind resistance distribution is a fixed value; perform unsupervised classification on the error e between the measured value and the simulated value of the fermentation stack parameter and the error between the measured value and the simulated value of the leakage position of the air supply duct in the historical data, determine the classification cluster Z to which the error e between the measured value and the simulated value of the fermentation stack parameter belongs, find the leakage position and the wind resistance at the leakage position corresponding to the error in the historical data in the classification cluster Z, determine the minimum value min1 and maximum value max1 of the leakage position from the starting point of the air supply duct, and the minimum value min2 and maximum value max2 of the wind resistance, and include the air supply duct whose distance from the starting point of the air supply duct is between min1 and max1 and whose wind resistance is between min2 and max2 into the key detection range.

[0061] First, the air supply ducts within the key inspection range are tested for leaks. If no leaks are detected, the air supply ducts at other locations are tested. If leaks are detected, the leaking locations are repaired.

[0062] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0063] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An intelligent multi-state perception ventilation control method based on simulation, characterized in that: The following steps are involved: S11, obtaining historical data of the fermentation stack, and according to whether there is a leak in the air supply pipeline, the historical data of the fermentation stack is divided into historical data when there is no leak and historical data when there is a leak; S12, based on the historical data when there is no leakage, training the parameter prediction model of the fermentation stack, including steps S21 and S22: S21, based on historical data when there is no leakage, determining the time T when the air supply operation affects the fermentation stack; Based on the historical data when there is no leakage, the dividing point c1 between air supply operation and no air supply operation and the dividing point c2 between no air supply operation and air supply operation are determined, and the parameter data between the dividing points c1 and c2 are used as the first input feature, and the parameter data between the dividing points c2 and c1 are used as the second input feature; the air supply operation and the no air supply operation are switched alternately, and there are several groups of first input features and second input features; For each group of first input features, supervised classification is used for classification. In a group of first input features, the 1st, 2nd, 3rd, ...T1+1th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The first classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T1+1, and the classification accuracy A1 of the negative examples is calculated by the following formula: , N21 is the number of negative examples in the group, N11 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT1 of T1 obtained according to the first input feature of all groups; For each group of second input features, supervised classification is used for classification. In a group of second input features, the 1st, 2nd, 3rd, ...T2th input features are used as positive examples, and the remaining input features in the group are used as negative examples. The second classification model is trained until the classification accuracy of the negative examples is less than the preset accuracy requirement. At this time, the number of positive examples is T2+1, and the classification accuracy A2 of the negative examples is calculated by the following formula: , N22 is the number of negative examples in the group, and N12 is the number of negative examples in the group classified as negative examples by the classification model; take the average value AT2 of T2 obtained according to the second input feature of all groups, round up the means of AT1 and AT2, and get the time T that the air supply operation affects the fermentation pile; S22, the historical data without leakage is divided into a training set and a test set. In the training set, all fermentation pile parameters, air supply parameters g, organic waste type type and fermentation time at time t and after time t are taken as input. As output, the For the i-th fermentation pile parameter, train the prediction model of the i-th fermentation pile parameter, and use the historical data of the test set to verify the prediction model of the i-th fermentation pile parameter; obtain the prediction model of all fermentation pile parameters; if air supply is in progress at time t, the air supply parameter g is 1, if air supply is not in progress at time t, the air supply parameter g is 0; distinguish the types of organic waste by assigning different discrete values ​​to type; S13, based on the parameter prediction model of the fermentation stack, verify the parameters of the fermentation stack to determine whether there is a leakage event in the air supply pipeline, including steps S41 to S43: S41, in the historical data when there is no leakage, the parameter data of T consecutive fermentation stacks, the air supply parameter g, the type of organic waste type and the fermentation time are input as input to the prediction model of the i-th fermentation stack parameter, and the i-th fermentation stack parameter simulation value is obtained, and the error between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter is calculated; there are K1 time windows in the historical data, and the error between K1 measured values ​​and the simulated values ​​is obtained; In the historical data when there is leakage, the parameter data of T consecutive fermentation piles, the air supply parameter g, the type of organic waste type and the fermentation time are used as inputs and input into the prediction model of the i-th fermentation pile parameter to obtain the i-th fermentation pile parameter simulation value, and calculate the error between the measured value and the simulated value of the T+1-th i-th fermentation pile parameter; there are K2 time windows in the historical data, and the errors between K2 measured values ​​and simulated values ​​are obtained; S42, obtaining all parameter data of the fermentation stack, inputting parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of the i-th fermentation stack parameter, obtaining the T+1-th i-th fermentation stack parameter simulation value, and calculating the error e between the measured value and the simulated value of the T+1-th i-th fermentation stack parameter; S43, calculate the leakage probability of the air supply duct , , is the leakage probability corresponding to the i-th fermentation pile parameter, , The number of historical data with or without leakage in which the error between the measured value and the simulated value is greater than or equal to e; is the number of historical data in which the error between the measured value and the simulated value is greater than or equal to e in the historical data when there is leakage; if the leakage probability of the air supply duct If the value is greater than or equal to the preset threshold, it is determined that there is a leakage event in the air supply duct; otherwise, it is determined that there is no leakage event in the air supply duct; If there is a leakage event, the air supply duct is leak tested; if there is no leakage event, the parameters of the fermentation stack are continuously verified.

2. The intelligent multi-state sensing ventilation control method based on simulation according to claim 1 is characterized in that: In step S13, the process of determining whether there is a leakage in the air supply duct further comprises the following steps: Obtain the error e between the measured value and the simulated value of the fermentation stack parameter; determine the leakage position of the air supply duct and the error between the measured value and the simulated value corresponding to the leakage position in the historical data when there is a leakage; determine the wind resistance distribution of the air supply duct according to the installation situation and material properties of the air supply duct, and the wind resistance distribution is a fixed value; perform unsupervised classification on the error e between the measured value and the simulated value of the fermentation stack parameter and the error between the measured value and the simulated value of the leakage position of the air supply duct in the historical data, determine the classification cluster Z to which the error e between the measured value and the simulated value of the fermentation stack parameter belongs, find the leakage position and the wind resistance at the leakage position corresponding to the error in the historical data in the classification cluster Z, determine the minimum value min1 and maximum value max1 of the leakage position from the starting point of the air supply duct, and the minimum value min2 and maximum value max2 of the wind resistance, and include the air supply duct whose distance from the starting point of the air supply duct is between min1 and max1 and whose wind resistance is between min2 and max2 into the key detection range.

3. The intelligent multi-state sensing ventilation control method based on simulation according to claim 2 is characterized in that: In step S13, the air supply duct is leak-checked, which further includes the following steps: First, the air supply ducts within the key inspection range are tested for leaks. If no leaks are detected, the air supply ducts at other locations are tested. If leaks are detected, the leaking locations are repaired.

4. An intelligent multi-state perception ventilation control system based on simulation, using an intelligent multi-state perception ventilation control method based on simulation as described in any one of claims 1-3, characterized in that: include: Data monitoring module, status analysis module, data storage module, detection module and maintenance management module; The output end of the data monitoring module is connected to the input end of the data storage module, and is used to obtain the parameter data of the fermentation stack and store it in the data storage module; the output end of the state analysis module is connected to the input end of the detection module, and the integrity of the air supply duct is confirmed according to the parameter data of the fermentation stack, the fermentation time and the air supply condition; the output end of the data storage module is connected to the input end of the state analysis module, and is used to store the parameter data and air supply condition data during the fermentation process; the output end of the detection module is connected to the input end of the maintenance management module, and is used to perform leakage detection on the air supply duct; the maintenance management module, when the detection module determines that there is a leakage in the air supply duct, contacts the management personnel to repair the air supply duct.

5. The intelligent multi-state sensing ventilation control system based on simulation according to claim 4 is characterized in that: The data monitoring module also includes a sensor unit, a timer unit and a measurement and transmission unit; the sensor unit is used to obtain parameter data of the fermentation pile; the timer unit is used to determine the duration of the fermentation process; the measurement and transmission unit is used to determine the air supply condition of the air supply duct.

6. The intelligent multi-state sensing ventilation control system based on simulation according to claim 4 is characterized in that: The state analysis module also includes an impact time determination unit, a fermentation stack parameter prediction unit, a simulation detection unit, a data dimension reduction unit and a leakage analysis unit; the impact time determination unit is used to determine the time window from the start of the air supply operation to the actual impact on the fermentation stack; the fermentation stack parameter prediction unit is used to train the prediction model of each fermentation stack parameter; the simulation detection unit is used to obtain the fermentation stack parameters in the absence of leakage; the leakage analysis unit determines the probability of a leakage event in the air supply duct based on the error between the simulated value and the measured value of the fermentation stack parameter; the data dimension reduction unit is used to perform dimensionality reduction processing on the data within the time window.

7. The intelligent multi-state sensing ventilation control system based on simulation according to claim 6 is characterized in that: The simulation detection unit inputs the parameter data of T consecutive fermentation stacks, air supply parameter g, organic waste type type and fermentation time into the prediction model of each fermentation stack parameter, obtains the simulated value of the T+1th fermentation stack parameter when there is no leakage, and compares it with the actual value of the T+1th fermentation stack parameter to determine the error between the simulated value and the measured value of the fermentation stack parameter.

Images