Intelligent treatment method for toilet odor source

Abstract

The embodiment of the invention relates to an intelligent treatment method for a toilet peculiar smell source, which comprises the following steps: firstly, acquiring real-time temperature, humidity, various gas concentrations and peculiar smell intensity from a temperature and humidity sensor, an ammonia gas sensor, a hydrogen sulfide sensor, a TVOC (Total Volatile Organic Compound) sensor and a peculiar smell sensor according to a preset frequency to form current monitoring data; then, whether all the gas concentrations exceed the threshold value or not is recognized, if yes, the force of the ventilation and exhaust device is reduced, and the ion generator and the photocatalytic oxidation module are closed; if not, the ventilation force is increased, the device is started, and the deodorant atomizing and spraying device is driven to spray. Then, whether the odor intensity exceeds a threshold value or not is judged, if yes, an environment monitoring vector is generated based on current data, an odor source recognition model is input to obtain a classification vector, the total probability of the sources exceeding the threshold value is counted, a manual cleaning instruction is sent to a cleaning client when the total probability is larger than 0, and precise treatment of the toilet odor is achieved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to an intelligent method for controlling odor sources in bathrooms. Background Technology

[0002] Unpleasant odors in restrooms are a widespread problem, seriously affecting user experience and health. Traditional methods of odor control have many drawbacks:

[0003] First, relying on regular manual inspections makes it impossible to monitor environmental parameters such as bathroom temperature, humidity, and concentration of harmful gases in real time and continuously. This makes it difficult to detect and deal with initial odors in a timely manner. Usually, the problem is only addressed when the odor significantly affects the use of the bathroom, resulting in the problem not being resolved in a timely manner.

[0004] Secondly, the approach to odor control lacks specificity. After detecting an odor, many simply increase ventilation or use general deodorizing products without considering the specific cause. In reality, bathroom odor sources are diverse, such as ammonia, hydrogen sulfide, and TVOCs. Different odor sources require different treatment strategies, and traditional methods are insufficient for precise control, leading to recurring odors.

[0005] Third, the existing monitoring equipment has fragmented functions, with sensors working independently and data lacking integrated analysis. This makes it impossible to provide comprehensive, accurate, and targeted decision-making basis for governance, and difficult to formulate the optimal governance plan.

[0006] With the development of technology, the demand for intelligent and precise environmental monitoring and management is becoming increasingly urgent. There is a pressing need for an intelligent method that can monitor and accurately identify odor sources in real time and address them in a targeted manner to improve the quality of the bathroom environment. Summary of the Invention

[0007] The purpose of this invention is to address the shortcomings of existing technologies by providing an intelligent method for controlling odor sources in bathrooms, thereby solving the problems existing in the prior art.

[0008] To achieve the above objectives, the present invention provides an intelligent method for controlling odor sources in bathrooms, the method comprising:

[0009] According to the preset first monitoring frequency, it periodically obtains a set of real-time temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity from the temperature and humidity sensor, ammonia sensor, hydrogen sulfide sensor, TVOC sensor and odor sensor connected to it, and combines them into a current monitoring data.

[0010] The system identifies whether the concentration of all gases in the current monitoring data does not exceed their respective concentration thresholds. If so, it drives the ventilation and exhaust device to reduce the ventilation intensity and shuts down the ion generator and photocatalytic oxidation module. If not, it drives the ventilation and exhaust device to increase the ventilation intensity, turns on the ion generator and photocatalytic oxidation module, and drives the deodorant atomizing spray device to complete the specified number of spray operations.

[0011] The system identifies whether the odor intensity of the current monitoring data exceeds the preset odor intensity threshold. If so, it generates a corresponding environmental monitoring vector based on the current monitoring data, inputs the current environmental monitoring vector into the odor source identification model to predict and obtain the corresponding odor source classification vector, and counts the total probability of the source whose probability value in the current odor source classification vector exceeds the preset probability threshold. When the total probability is greater than 0, the system sends the preset manual cleaning instruction for the toilet to the cleaning client.

[0012] In one possible implementation, the method further includes the following preceding steps:

[0013] Temperature and humidity sensors, various gas sensors, odor sensors, intelligent detection equipment, and various treatment devices are deployed in the bathroom;

[0014] The various gas sensors include ammonia sensors, hydrogen sulfide sensors, and TVOC sensors; the various treatment devices include: ventilation and exhaust devices, ion generators, photocatalytic oxidation modules, and deodorant atomizing spray devices.

[0015] In one possible implementation, the temperature and humidity sensor is used to monitor the temperature and humidity of the bathroom in real time; the ammonia sensor is used to monitor the ammonia concentration of the bathroom in real time; the hydrogen sulfide sensor is used to monitor the hydrogen sulfide concentration of the bathroom in real time; the TVOC sensor is used to monitor the total volatile organic compound concentration of the bathroom in real time; and the odor sensor is used to simulate the human olfactory system, utilizing the interaction between a sensitive material and some or all of the above eight types of gas molecules to detect the overall odor intensity.

[0016] In one possible implementation, the method further includes, prior to: training an odor source identification model;

[0017] The input to the odor source identification model is an environmental monitoring vector, which includes temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity.

[0018] The output of the odor source identification model is an odor source classification vector, which consists of N source probabilities y. i Composition; N is the preset total number of odor source objects.

[0019] In one possible implementation, the odor source identification model includes a feature extraction network, a fully connected layer, and a sigmoid function layer connected sequentially.

[0020] The feature extraction network is implemented based on a CNN model or an MLP model and is used to perform feature encoding processing on the environmental monitoring vector X to obtain the corresponding encoded feature H1, which is then sent to the fully connected layer.

[0021] The fully connected layer is used to transform the encoded feature H1 into odor source features to obtain the corresponding feature vector H2, which is then sent to the softmax function layer; wherein, the feature vector H2 has a length of N, consisting of N vector data h i , 1≤i≤N;

[0022] The sigmoid function layer is used to calculate the vector data h of each feature vector H2. i The corresponding source probability y is obtained by performing probability calculation. i ; and the obtained N source probabilities y i The corresponding odor source classification vector Y is output; where the source probability y is... i The calculation method is as follows:

[0023] In one possible implementation, the odor source identification model dataset consists of multiple first data records; the first data records include training vector X. tr and label vector Y * ;

[0024] The training sequence X tr The data format is consistent with the environmental monitoring vector X;

[0025] The label vector Y * The data format is consistent with the odor source classification vector Y; the label vector Y * The source probability y in * i The value of y is either 0 or 1; there are N source probabilities y * i All values ​​of 0 indicate the absence of a significant odor source; the probability y of N sources is... * i If some or all of them are 1, it indicates the existence of one or more obvious sources of odor.

[0026] In one possible implementation, the model dataset is collected in the following ways:

[0027] During the data collection process, multiple data collections are performed on the temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity inside the bathroom using temperature and humidity sensors, ammonia sensors, hydrogen sulfide sensors, TVOC sensors, and odor sensors to generate corresponding multiple sets of data collection arrays. Each set of data collection arrays includes temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity.

[0028] Based on the set of odor source objects, one or more types of odor source objects in the bathroom are calibrated to obtain a corresponding odor source calibration record; and a corresponding training vector X is generated based on each collection array. tr A corresponding label vector Y is generated based on the odor source calibration records corresponding to each acquisition array. * ; and from each training vector X tr and its corresponding label vector Y * This forms a corresponding first data record;

[0029] The model dataset is composed of the first data records generated by multiple data collectors.

[0030] In one possible implementation, the training method for the odor source identification model includes:

[0031] The model dataset is divided into two subsets according to a predetermined ratio: a training set and an evaluation set.

[0032] The training sequences X of the training set tr Input the odor source identification model for prediction, and use the odor source classification vector Y obtained from this prediction as the corresponding prediction vector Y. pre Then, based on each prediction vector Y... pre Its corresponding label vector Y * Form a prediction-label pair (Y pre ,Y * );

[0033] All prediction-label pairs (Y) corresponding to the training set pre ,Y * Substitute into the model loss function L M The first loss value is obtained through calculation; where the model loss function L M Implemented based on the cross-entropy loss function;

[0034] Identify whether the first loss value meets the preset first loss value range;

[0035] If this is not satisfied, the preset model optimizer is used to optimize the model loss function L. MThe model parameters of the odor source identification model are optimized in one round to reach the minimum value; and at the end of this round of optimization, the training sequences X of the training set are returned. tr Steps for inputting odor source identification models for prediction;

[0036] If satisfied, evaluate each training sequence X in the evaluation set. tr Input the odor source identification model for prediction, and use the odor source classification vector Y obtained from this prediction as the corresponding prediction vector Y. pre Then, based on each prediction vector Y... pre Its corresponding label vector Y * Form a prediction-label pair (Y pre ,Y * );

[0037] Based on all prediction-label pairs (Y) corresponding to the evaluation set pre ,Y * The F1 score is calculated to obtain the corresponding first F1 score;

[0038] The system identifies whether the first F1 value meets the preset range of first F1 values; if it does not meet the range, it returns 1; if it does meet the range, it stops training.

[0039] Secondly, the present invention provides a chip system including a processor coupled to a memory, the memory storing program instructions, wherein when the program instructions stored in the memory are executed by the processor, the intelligent management method for bathroom odor sources based on log serialization as described in any of the first aspects is implemented.

[0040] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor using the intelligent treatment method for odor sources in a bathroom as described in any one of the first aspects.

[0041] The intelligent method for controlling odor sources in bathrooms provided by this invention has the following technical effects:

[0042] A complete intelligent management system is formed by deploying multiple sensors, intelligent detection devices, and treatment devices in the bathroom. All parts collaborate and share data. The intelligent detection devices precisely control the operation of the treatment devices, overcoming the limitations of traditional decentralized equipment and improving the overall management level. Real-time data is acquired from multiple sensors at preset frequencies to form monitoring data, overcoming the lag of manual inspections. It can promptly capture changes in bathroom temperature, humidity, and harmful gas concentrations, detecting odors in their early stages and providing timely and accurate data support for treatment, preventing the spread of odors.

[0043] Treatment is tiered based on whether the gas concentration exceeds the threshold. If none exceed the threshold, ventilation is reduced and some devices are shut down for energy saving; if the threshold is exceeded, ventilation is increased and devices such as ion generators are turned on, while deodorizing agents are sprayed. This avoids the indiscriminate nature of traditional treatment methods, improves targeting and efficiency, and reduces costs.

[0044] When the odor intensity exceeds a threshold, an environmental monitoring vector is generated and input into the odor source identification model to obtain a classification vector. The total probability of sources exceeding the threshold is counted, and a cleaning command is sent when the probability is greater than 0. This system can accurately identify odor sources, such as ammonia and TVOC, allowing cleaning staff to target specific sources and solve the problem of incomplete treatment.

[0045] A model for identifying odor sources is trained by taking a vector containing multiple environmental parameters as input and outputting a classification vector composed of source probabilities. The model employs a specific structure and, through training on a large dataset, can continuously learn and optimize, improving the accuracy of identifying different odor sources and providing a basis for precise odor control. Attached Figure Description

[0046] Figure 1 This is a structural diagram of an intelligent system for controlling odor sources in bathrooms.

[0047] Figure 2 This is a schematic diagram of the intelligent odor control method for bathrooms provided in an embodiment of the present invention;

[0048] Figure 3 This is a structural diagram of an odor source identification model. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0050] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0051] Example 1

[0052] Figure 1 Here is a structural diagram of an intelligent system for controlling odor sources in bathrooms, such as... Figure 1 As shown, the following will be deployed in the restroom: temperature and humidity sensors, multiple types of gas sensors, odor sensors, intelligent detection equipment, and multiple treatment devices; in addition, a corresponding cleaning client will be provided for the restroom cleaning staff.

[0053] Various types of gas sensors include: ammonia sensors, hydrogen sulfide sensors, and TVOC (Total Volatile Organic Compounds) sensors;

[0054] Various treatment devices include: ventilation and exhaust systems, ion generators, photocatalytic oxidation modules, and deodorizing agent atomizing spray devices.

[0055] Temperature and humidity sensors are used to monitor the temperature and humidity in the bathroom in real time. Ammonia sensors are used to monitor the ammonia concentration in the bathroom in real time. Hydrogen sulfide sensors are used to monitor the hydrogen sulfide concentration in the bathroom in real time. TVOC sensors are used to monitor the total volatile organic compound concentration in the bathroom in real time. Under normal circumstances, eight gaseous components are defined for odorous and foul-smelling substances, including: NH3 (ammonia), H2S (hydrogen sulfide), CH4S (methanethiol), C2H6S (methyl sulfide), CS2 (carbon disulfide), C2H6S2 (dimethyl disulfide), C3H9N (trimethylamine), and C8H8 (styrene).

[0056] Odor sensors simulate the human olfactory system, using sensitive materials to detect overall odor intensity by interacting with some or all of the eight types of gas molecules mentioned above. The treatment system includes a ventilation and exhaust system, an ion generator, and a deodorizing agent atomizing spray device. The ventilation and exhaust system is used to ventilate and exchange air in the bathroom. The ventilation and air exchange intensity can be increased or decreased by adjusting this device.

[0057] The ion generator releases negative ions to neutralize odor molecules. The photocatalytic oxidation module uses ultraviolet (UV) light to irradiate a photocatalyst (such as TiO2), exciting the surface of the photocatalyst to generate electron-hole pairs. High-energy holes possess strong oxidizing properties, directly oxidizing odor molecules (such as volatile organic compounds, VOCs) adsorbed on the catalyst surface, decomposing them into harmless CO2 and H2O. Simultaneously, electrons combine with oxygen in the air to generate superoxide anion radicals, and holes react with water to generate hydroxyl radicals. These active radicals further undergo a chain reaction with odor molecules, completely mineralizing them into non-toxic small molecules.

[0058] The deodorizing agent atomizing spray device achieves the deodorizing effect by spraying bio-enzymes or nano-deodorizing agents.

[0059] Figure 2 This is a schematic diagram of the intelligent odor control method for bathrooms provided in an embodiment of the present invention. Figure 2 As shown, the method includes the following steps:

[0060] Step 210: Periodically obtain a set of real-time temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity from the temperature and humidity sensor, ammonia sensor, hydrogen sulfide sensor, TVOC sensor and odor sensor connected to itself at a preset first monitoring frequency, and form a current monitoring data.

[0061] Step 220: Identify whether the concentration of all gases in the current monitoring data does not exceed their respective concentration thresholds; if so, drive the ventilation and exhaust device to reduce the ventilation intensity and turn off the ion generator and photocatalytic oxidation module; if not, drive the ventilation and exhaust device to increase the ventilation intensity, turn on the ion generator and photocatalytic oxidation module to work, and drive the deodorant atomizing spray device to complete the specified number of spray operations.

[0062] Step 230: Identify whether the odor intensity of the current monitoring data exceeds the preset odor intensity threshold; if so, generate a corresponding environmental monitoring vector based on the current monitoring data, input the current environmental monitoring vector into the odor source identification model to predict and obtain the corresponding odor source classification vector, and count the total probability of the source probability that exceeds the preset probability threshold in the current odor source classification vector, and send the preset toilet manual cleaning instruction to the cleaning client when the total probability is greater than 0.

[0063] The following is a detailed explanation of the odor source identification model:

[0064] The odor source identification model is used to classify and identify the gas sources that produce odors in the bathroom based on the environmental monitoring vector X input to the model, and outputs the corresponding odor source classification vector Y. After training, the odor source identification model is installed on intelligent detection equipment for intelligent detection.

[0065] The model input is an environmental monitoring vector X, which includes: temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity.

[0066] The model output is an odor source classification vector Y, consisting of N source probabilities y i Composition, where N is the preset total number of odor source objects.

[0067] The predefined set of odor source objects in this application consists of N types of odor source objects, such as urine, feces, and decaying garbage. The source probability y... i Each odor source object corresponds one-to-one with the odor source object set.

[0068] like Figure 3 As shown, the odor source identification model consists of a feature extraction network, a fully connected layer, and a sigmoid function layer connected sequentially.

[0069] The feature extraction network is implemented based on a CNN model or an MLP model. It is used to perform feature encoding on the environmental monitoring vector X to obtain the corresponding encoded features H1, which are then sent to the fully connected layer.

[0070] The fully connected layer is used to transform the encoded feature H1 into odor source features, obtaining the corresponding feature vector H2, which is then sent to the softmax function layer. The feature vector H2 has a length of N and consists of N vector data h. i , 1≤i≤N.

[0071] The sigmoid function layer is used to calculate the vector data h of the eigenvector H2. i The corresponding source probability y is obtained by performing probability calculation. i ; and the obtained N source probabilities y i This forms the corresponding odor source classification vector Y, which is output. Here, the source probability y... i The calculation method is as follows:

[0072] The data format of the model dataset is as follows: The model dataset consists of multiple first data records; each first data record includes the training vector X. tr and label vector Y * Training sequence X tr The data format is consistent with the environmental monitoring vector X. Label vector Y * The data format is consistent with the odor source classification vector Y; the label vector Y * The source probability y in * i The value of y is either 0 or 1; there are N source probabilities y * i A value of 0 for all values ​​indicates the absence of a significant odor source; the probability y of N sources is... * i A value of 1 in some or all of the values ​​indicates the existence of one or more distinct sources of odor.

[0073] The data acquisition methods for the odor source identification model include:

[0074] 1. Recruit a large number of data collectors to collect data from multiple restrooms.

[0075] 2. During the data collection process, the data collector used temperature and humidity sensors, ammonia sensors, hydrogen sulfide sensors, TVOC sensors, and odor sensors to collect data multiple times on the temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity inside the bathroom, generating multiple sets of data collection arrays (temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration, and odor intensity). Based on the odor source object set, the main one or more types of odor sources in the bathroom were calibrated to obtain a corresponding odor source calibration record. A corresponding training vector X was generated based on each data collection array. tr And based on the odor source calibration records corresponding to each collection array, a corresponding label vector Y is generated. * ; and from each training vector X tr and its corresponding label vector Y * This forms a corresponding first data record.

[0076] 3. The model dataset consists of the first data records generated by all data collectors.

[0077] The training method for the odor source identification model is as follows:

[0078] 1. Divide the model dataset into two subsets according to a predetermined ratio: training set and evaluation set.

[0079] 2. Combine the training sequences X in the training set. tr Input the odor source identification model for prediction, and use the odor source classification vector Y obtained from this prediction as the corresponding prediction vector Y. pre Then, based on each prediction vector Y... pre Its corresponding label vector Y * Form a prediction-label pair (Y pre ,Y * ).

[0080] 3. Combine all prediction-label pairs (Y) corresponding to the training set. pre ,Y * Substitute into the model loss function L M The first loss value is obtained through calculation.

[0081] Wherein, the model loss function L M Implemented based on the cross-entropy loss function.

[0082] 4. Identify whether the first loss value meets the preset first loss value range; if it does not meet the range, proceed to step 5; if it does meet the range, proceed to step 6.

[0083] 5. Use a preset model optimizer (Adam optimizer, SGD optimizer, etc.) to optimize the model loss function L. MThe direction that reaches the minimum value is used to optimize the model parameters of the odor source identification model in one round; and 2 is returned at the end of this round of optimization.

[0084] 6. Assign each training sequence X to the evaluation set. tr Input the odor source identification model for prediction, and use the odor source classification vector Y obtained from this prediction as the corresponding prediction vector Y. pre Then, based on each prediction vector Y... pre Its corresponding label vector Y * Form a prediction-label pair (Y pre ,Y * ).

[0085] 7. Based on all prediction-label pairs (Y) corresponding to the evaluation set pre ,Y * The F1 value is calculated to obtain the corresponding first F1 value.

[0086] The system identifies whether the first F1 value meets the preset range of first F1 values; if it does not meet the range, it returns 1; if it does meet the range, it stops training.

[0087] The intelligent method for controlling odor sources in bathrooms provided by this invention has the following technical effects:

[0088] A complete intelligent management system is formed by deploying multiple sensors, intelligent detection devices, and treatment devices in the bathroom. All parts collaborate and share data. The intelligent detection devices precisely control the operation of the treatment devices, overcoming the limitations of traditional decentralized equipment and improving the overall management level. Real-time data is acquired from multiple sensors at preset frequencies to form monitoring data, overcoming the lag of manual inspections. It can promptly capture changes in bathroom temperature, humidity, and harmful gas concentrations, detecting odors in their early stages and providing timely and accurate data support for treatment, preventing the spread of odors.

[0089] Treatment is tiered based on whether the gas concentration exceeds the threshold. If none exceed the threshold, ventilation is reduced and some devices are shut down for energy saving; if the threshold is exceeded, ventilation is increased and devices such as ion generators are turned on, while deodorizing agents are sprayed. This avoids the indiscriminate nature of traditional treatment methods, improves targeting and efficiency, and reduces costs.

[0090] When the odor intensity exceeds a threshold, an environmental monitoring vector is generated and input into the odor source identification model to obtain a classification vector. The total probability of sources exceeding the threshold is counted, and a cleaning command is sent when the probability is greater than 0. This system can accurately identify odor sources, such as ammonia and TVOC, allowing cleaning staff to target specific sources and solve the problem of incomplete treatment.

[0091] A model for identifying odor sources is trained by taking a vector containing multiple environmental parameters as input and outputting a classification vector composed of source probabilities. The model employs a specific structure and, through training on a large dataset, can continuously learn and optimize, improving the accuracy of identifying different odor sources and providing a basis for precise odor control.

[0092] Example 2

[0093] Embodiment 2 of the present invention provides a computer server, including: a memory, a processor, and a transceiver;

[0094] The processor is used to couple with the memory, read and execute instructions in the memory to implement any of the intelligent treatment methods for bathroom odor sources provided in Embodiment 1 above;

[0095] The transceiver is coupled to the processor, and the processor controls the transceiver to send and receive messages.

[0096] Example 3

[0097] Embodiment 3 of the present invention provides a chip system, including a processor and a memory coupled together. The memory stores program instructions. When the program instructions stored in the memory are executed by the processor, any one of the intelligent treatment methods for bathroom odor sources provided in Embodiment 1 is implemented.

[0098] Example 4

[0099] Embodiment 4 of the present invention provides a computer-readable storage medium including a program or instructions, which, when executed on a computer, implements the intelligent treatment method for bathroom odor sources as provided in Embodiment 1.

[0100] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0101] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented in hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0102] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for intelligent management of a source of bathroom odor, characterized in that, The method comprises: A set of real-time temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity is obtained from the temperature and humidity sensor, ammonia sensor, hydrogen sulfide sensor, TVOC sensor and odor sensor connected with itself at a preset first monitoring frequency, and a current monitoring data is formed; Whether all the gas concentrations of the current monitoring data exceed the corresponding concentration threshold is identified; if yes, the ventilation and exhaust device is driven to reduce the ventilation and exhaust intensity, and the ion generator and the photocatalytic oxidation module are turned off; if not, the ventilation and exhaust device is driven to increase the ventilation and exhaust intensity, the ion generator and the photocatalytic oxidation module are turned on and work, and the deodorant atomization spraying device is driven to complete a specified number of spraying operations; Whether the odor intensity of the current monitoring data exceeds the preset odor intensity threshold is identified; if yes, a corresponding environmental monitoring vector is generated based on the current monitoring data, the current environmental monitoring vector is input into the odor source identification model for prediction to obtain a corresponding odor source classification vector, and the total number of source probabilities whose probability values exceed the preset probability threshold in the current odor source classification vector is counted, and the preset toilet manual cleaning instruction is sent to the cleaning client when the total number of probabilities is greater than 0.

2. The method of claim 1, wherein, The method further comprises: A temperature and humidity sensor, a plurality of gas sensors, an odor sensor, an intelligent detection device and a plurality of treatment devices are deployed in the toilet. The plurality of gas sensors include an ammonia sensor, a hydrogen sulfide sensor and a TVOC sensor; the plurality of treatment devices include a ventilation and exhaust device, an ion generator, a photocatalytic oxidation module and a deodorant atomization spraying device.

3. The method of claim 2, wherein, The temperature and humidity sensor is used to monitor the temperature and humidity of the toilet in real time; the ammonia sensor is used to monitor the ammonia concentration of the toilet in real time; the hydrogen sulfide sensor is used to monitor the hydrogen sulfide concentration of the toilet in real time; the TVOC sensor is used to monitor the total volatile organic compound concentration of the toilet in real time; and the odor sensor is used to simulate the human olfactory system, and interact with part or all of the eight types of gas molecules to detect the overall odor intensity.

4. The method of claim 1, wherein, The method further comprises training an odor source identification model; The input of the odor source identification model is an environmental monitoring vector, which includes temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity; The output of the odor source identification model is an odor source classification vector, which is composed of N source probabilities y i ; N is the total number of preset odor gas source objects.

5. The method of claim 1, wherein, The odor source identification model comprises a feature extraction network, a full connection layer and a sigmoid function layer connected in sequence; The feature extraction network is realized based on a CNN model or an MLP model, and is used to perform feature coding processing on the environmental monitoring vector X to obtain corresponding coding features H1 and send them to the full connection layer; The full connection layer is used for odor source feature conversion on the coded features H1 to obtain a corresponding feature vector H2 which is sent to a softmax function layer; wherein the vector length of the feature vector H2 is N, and the feature vector H2 is composed of N vector data h i , 1≤i≤N. The sigmoid function layer is used to obtain each vector data h of the feature vector H2 i The corresponding source probability y is obtained by performing probability calculation i ; and the N source probabilities y i are combined to form a corresponding off-flavor source classification vector Y output; wherein the source probability y i is calculated in the following manner:

6. The method of claim 1, wherein, The odor source identification model model dataset is composed of a plurality of first data records; the first data records include a training vector X tr and a label vector Y * ; The training sequence X tr The data format is consistent with the environmental monitoring vector X; The label vector Y * The data format is consistent with the odor source classification vector Y; the label vector Y * The source probability The value in the label vector Y N source probabilities are all 0, indicating that there is no obvious odor source object; N source probabilities Part or all are 1, indicating that there is one or more obvious odor source objects.

7. The method of claim 1, wherein, The model data set is collected in the following way: In the collection process, the temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity in the toilet are collected by the temperature and humidity sensor, ammonia sensor, hydrogen sulfide sensor, TVOC sensor and odor sensor to generate corresponding multiple sets of collection arrays, each set of the multiple sets of collection arrays including temperature, humidity, ammonia concentration, hydrogen sulfide concentration, TVOC concentration and odor intensity. Based on the odor gas source object set, one or more types of odor gas source objects in the toilet are calibrated to obtain a corresponding odor gas source calibration record; and based on each collection array, a corresponding training vector X is generated tr ; based on the corresponding odor gas source calibration record of each collection array, a corresponding label vector Y is generated * ; and each training vector X tr and its corresponding label vector Y * comprise a corresponding first data record; The first data records generated by the multiple collectors constitute a model data set.

8. The method of claim 1, wherein, The training method of the odor source identification model comprises: The model data set is divided into two sub-data sets, a training set and an evaluation set, according to a predetermined proportion. Each training sequence X tr The odor source recognition model is inputted for prediction, and the odor source classification vector Y obtained by this time prediction is taken as the corresponding prediction vector Y pre Each prediction vector Y pre is taken as the corresponding label vector Y * A prediction-label pair (Y pre ,Y * ) is formed. all the prediction-label pairs (Y pre ,Y * ) corresponding to the training set are brought into a model loss function L M to calculate a first loss value; wherein the model loss function L M is realized based on a cross-entropy loss function. It is identified whether the first loss value meets a preset first loss value range. If not satisfied, use the preset model optimizer to optimize the model parameters of the odor source identification model in the direction of minimizing the model loss function L M to the minimum; and return the step of inputting each training sequence X tr of the training set into the odor source identification model for prediction at the end of the current optimization. If satisfied, each training sequence X of the evaluation set is evaluated tr The odor source identification model is input for prediction, and the odor source classification vector Y obtained by this prediction is taken as the corresponding prediction vector Y pre ; and each prediction vector Y pre and its corresponding label vector Y * comprise a prediction-label pair (Y pre , Y * ). performing F1 value calculation on all prediction-label pairs (Y pre ,Y * ) corresponding to the evaluation set to obtain a corresponding first F1 value; It is identified whether the first F1 value meets a preset first F1 value range; if not, the step of dividing the model data set into two sub-data sets according to a predetermined proportion is returned; if yes, the training is stopped.

9. A chip system, characterized by The processor is coupled with a memory, and the memory stores program instructions, which, when executed by the processor, implement the intelligent management method of the toilet odor source based on log serialization according to any one of claims 1-8.

10. A computer-readable storage medium, characterized in that, The computer program is stored on the computer readable storage medium, and the computer program is executed by the processor to implement the intelligent management method of the toilet odor source according to any one of claims 1-8.

Images