Traditional agricultural tools are not only the prototypes for modern agricultural machinery but also carriers of our agrarian culture and Chinese civilization. The aesthetic of traditional agricultural tools is a cultural blend of art and reality, society and nature, encapsulating the interconnected relationships between people and objects, objects with other objects, objects with their environment, and objects with society. This article takes the plow, the seedling horse, and the waterwheel, three representative traditional agricultural tools, as examples to delve into the functional, stylistic, manufacturing techniques, and cultural significance in the aesthetic of traditional Chinese agricultural tools. Their functional and structural designs fully embody the concept of humancenteredness, possessing various practical functions. In terms of form, they reflect the philosophical thought of harmony with nature, aligning with the aesthetic values of the East, adding beauty and interest to agricultural labor. Their craftsmanship is both aesthetically pleasing and practical, intricate yet reliable, showcasing the ingenuity of ancient craftsmanship. Consequently, they have become important subjects in ancient rituals, poetry, and artistic works, possessing profound historical and cultural value.Taking the example of supporting rural revitalization, the research explores the contemporary applications of the aesthetic of traditional agricultural tools and their important roles in preserving rural history and culture, stimulating the development of the cultural tourism industry, and promoting economic growth. These traditional agricultural tools have significant supportive roles, practical value, and realworld significance in these contexts.

Agricultural green technology is the critical element influencing the implementation effect of agricultural green development. On account of the bibliometric method  and traditional literature sorting method, and based on 996 core documents and CSSCI documents included in CNKI database from 1993 to 2022, this paper combs and analyzes the progress, hot spots and hot topic evolution in domestic agricultural green technology research field through CiteSpace software, and then puts forward research prospects. The results show  that the research and development stage in this field is clear, and the backbone has not been formed yet. There is still much room for progress in the cooperation network and external radiation capacity of research institutions. At this stage, the main research focuses on lowcarbon agriculture and circular agriculture, technology adoption and its influencing factors, technology innovation and policy frontier. Therefore, it may become the future research direction to closely follow the relevant policies of national agricultural green development to carry out in-depth research on agricultural green technology, deepen the research on agricultural green technology promotion mechanism and influencing elements, and strengthen the research on the effects and effects of technology acceptance.

 From the perspective of rural population aging, this article examines whether agricultural machinery has a substitutive effect on agricultural labor force. To determine whether there is a current shortage of agricultural labor in China, predict the future level of agricultural mechanization in China, and attempt to answer the question “Can agricultural mechanization solve the problem of agricultural labor shortage?”. Based on the provincial panel data of 31 provinces and cities in China from 2004 to 2020, from the perspective of the substitution ability of agricultural machinery for agricultural labor force, this paper uses the interactive fixed effect model to propose an estimation method to measure the substitution ability of agricultural machinery. The empirical results show that the substitution ability of agricultural machinery for agricultural labor force in China is as follows: 10000 kilowatt agricultural machinery can replace about 1169-1239 agricultural labor force every year. After that, the number of agricultural surplus labor force and the level of agricultural mechanization in China are predicted by the classical economic calculation method, time deductive reasoning method and ARIMA prediction model. The results show that the total power level of agricultural machinery in China is increasing at a rate of approximately 24946630 kilowatts per year. By 2035, the total power of agricultural machinery in China will increase by 374 199 450 kilowatts. The growth of agricultural mechanization can make up for the shortage of agricultural labor caused by population aging.

The implementation of the rural revitalization strategy has put forward higher requirements for farmers to get rich and increase their income. This paper discusses the problem of farmers' low income and difficulty in increasing their income, and explores the formulation of income increase strategies according to local conditions to clarify the way for farmers to get rich and increase their income. By analyzing the characteristics of farmers' income in Lupu Town of Yancheng City, this paper investigates and studies the current measures to increase farmers' income by developing efficient value-added agriculture, cultivating the economy of capable people, building high-standard farmland and cultivating excellent guides. Under the circumstances of weak industrial and commercial economic foundation and restricted development process of efficient agriculture in this region, this paper objectively puts forward strategies to increase farmers' income by clarifying the main line of rural revitalization development, implementing advantageous natural resources measures, anchoring industrial objectives, strengthening cadre team construction and other aspects, so as to provide ideas and theoretical support for the implementation of rural revitalization strategy to increase farmers' income.

In order to clarify the performance of lowcarbon technology adoption behavior and promote farmers' participation in low-carbon construction, based on 695 rice farmer survey data, data envelopment analysis and propensity score matching methods were comprehensively used to calculate the performance of farmers' low-carbon production adoption behavior and group differences. The results indicated that the average production performance of the sample farmers was 0.777 4, and there was still room for an increase of 0.222 6 to eliminate the impact of production factor mismatch. The impact of lowcarbon agricultural production technology on the overall agricultural production performance of sample farmers was positive, but not significant. The impact of lowcarbon agricultural production technology adoption by different types of farmers on agricultural production performance showed heterogeneity. The adoption of technology by three types of farmers such as life type, survival type and production type, had no significant impact on agricultural production performance. Only functional farmers could significantly promote the improvement of agricultural production performance. The research put forward policy recommendations such as strengthening publicity education and technology training, building a collaborative compensation mechanism between the government and the market for agricultural low-carbon transformation, implementing incentive policies and supporting measures based on the heterogeneity of farmers' business objectives.

In order to provide theoretical and practical basis for the design and innovative practice of ecological agriculture training model in the future, and then provide policy inspiration for high-quality agricultural development, based on the relevant statistical data of 31 provinces, cities and autonomous regions in China from 2012 to 2021, the coupling and coordination relationship among the four factors of policy support, market competition, environmental awareness and training resources was analyzed. In this study, the coupled coordination degree model was used to construct the evaluation model of the coordination degree affecting the factor system of ecological agriculture training effect，and the spatiotemporal evolution characteristics of the coupling coordination among the four factors of policy support, market competition, environmental awareness and training resources were analyzed. On this basis, the key factors limiting the ecological agriculture training effect were identified through the barrier degree model, and the changing trends and constraints of the coupling coordination of the elements affecting the ecological agriculture training effect in China were explored. The results showed that the coupling coordination degree of the four factors in the 31 provinces from 2012 to 2021, cities and autonomous regions of China was in an overall rising trend from 0.478 to 0.565, with an increase of 36.61%, but the coordination level was still not high, and it was in a state of barely coordinated development for a long time. In terms of spatial distribution, the coupling coordination degree of the four factors showed significant positive correlation among provinces and autonomous regions. The obstacle degree of each factor in descending order was as follows: market competition, training resources, policy support, environmental awareness, and the main obstacle factor affecting the coupling and coordination of various factors was the amount of government subsidies in policy support, the number of agricultural colleges in market competition, the amount of organic fertilizer input in environmental awareness and the area of the training venue in the training resources.

Under the background of lowcarbon cycle development, the value realization of carbon sink ecological products of agriculture, the countryside and farmers can promote the lowcarbon development of agriculture, the countryside and farmers, and also benefit farmers. Taking the household biogas project in the four prefectures of southern Xinjiang in the ecologically fragile area as the research object, and based on the perspective of low-carbon cycle, the small-scale methodology AMS-Ⅲ.R and AMS-Ⅰ.C were adopted to comprehensively calculate the dual carbon emission reduction benefits of greenhouse gases, estimate the economic benefits after participating in the CERs project d, and calculate the carbon emission reduction costs of the project, and finally the actual suggestions aiming at the actual regional situation were put forward. The results showed that in 2020, the total carbon emission reduction of household biogas project in the four prefectures of southern Xinjiang was 32.46×104 t/a, of which the first carbon emission reduction was 0.96×104 t/a, the second carbon emission reduction was 31.50×104 t/a, and the average carbon emission reduction per household was 1.12 t/a, the total economic income of farmers and herders participating in the CERs project was 33.85 million yuan, and the net present value of each household after participating in the CERs project was 2 229.6 yuan, the internal rate of return could reach 30% and the dynamic payback period was 3.16 years, the incremental cost per unit of carbon reduction (ICER) was 9.01 yuan/ton CO2 equivalent, which was ＄1.42/ton CO2 equivalent. Therefore, the development of the household biogas CERs project is feasible and has great development potential, and participating the carbon trading market to realize the value as one type of carbon sink products of agriculture, the countryside and farmers will also promote regional ecological restoration, carbon reduction and income increase, and consolidate the achievements of poverty alleviation.

During the upgrading of the structure of the domestic consumption and the strategic adjustment of agricultural structure, China's animal husbandry industry has developed rapidly, which leads to increasing of the demand for forage. However, the domestic supply of forage is insufficient and the safety problem is prominent. In this paper, the field investigations were conducted in Hangjinhou Banner and Dengkou county of Bayannur city, the pilot county of “grain transformation to feed”, to understand the input and output of local farmers' forage planting, the empirical research was carried out on  the influencing factors of farmers forage planting behavior by using the Logit, 2SLS and LIML models. The results showed that the forage grass planted by farmers in the survey area mainly included silage corn, alfalfa and oat grass, and the economic benefit of planting alfalfa grass was higher than other forage materials, with the net income of 22005 yuan per hectare in 2021, and higher than 4590 yuan per hectare and 12180 yuan per hectare compared with silage corn and oat grass respectively. However, the survey also found that the enthusiasm of farmers to plant forage, especially alfalfa, was not high. Further study on the influencing factors of farmers' choice of planting forage found that policy incentives, land endowment and organizational system factors had significant positive effects on farmers' behavior of planting forage. It is proposed that the government should establish a perfect land transfer market system as soon as possible to help farmers expand the scale of planting, smoothly carry out the structural adjustment of planting, increase the area of forage planting, promote the virtuous cycle of planting for breeding, solve the problem of shortage of forage in the breeding industry, and encourage all kinds of cooperatives to play the advantages of the cooperative system, help farmers reduce the input cost of planting forage, broaden the sales channels of forage, and increase the income of farmers. Countermeasures and suggestions to increase farmers' income.

In order to improve the accuracy of maize disease leaf recognition model, an improved YOLOv3 maize disease recognition method was proposed. First of all, in order to obtain deeper maize disease characteristics, the YOLOv3 network architecture was modified to YOLOv3-M1 and YOLOv3-M2 by changing the proportion of shallow feature map and adding a fourth detection layer. Then, the improved K-means algorithm was used for clustering, and the obtained anchor frame tended to be the true boundary frame of the data set. Finally, a balance factor was added for each category, and the difficulty of samples in different categories was weighted to modify the loss function, so that the model could find the best point between the boundary box prediction and the category prediction, so that the algorithm could obtain the best detection effect. The test results show that the accuracy of the improved YOLOv3-M1 and YOLOv3-M2 models in the test set is as high as 95.63% and 97.59%, respectively. Compared with the YOLOv3 model, the recognition accuracy is increased by 4.15% and 6.28%, respectively, and the recognition accuracy is greatly improved in the corn data set.

In order to satisfy the automatic driving function of the boat tractor, this paper designed a set of YOLOv5 integrated SGBM algorithm machine vision obstacle perception system. Firstly, people, machinetiller and farm tools were taken as objects to shoot and collect images to get paddy field obstacle data set. The images were input into the YOLOv5 network model for iterative training to get the optimal weight. Later, the most weight was used for testing and compared with YOLOv4 and Faster R-CNN networks. The left and right images taken by the binocular camera were input into the YOLOv5 model for detection. After correcting and transforming the output information of the target obstacle detection box, the SGBM algorithm was used for parallax calculation to complete the target obstacle recognition and depth estimation. The results show that the average accuracy of YOLOv5 is stable at 87.25%, 1.55% higher than that of YOLOv4, 4.04% higher than that of Faster R-CNN, and the detection time of a single image is 0.017 s, 0.081 s faster than that of YOLOv4. It is 0.182 s faster than Faster R-CNN, and the model size is only 13.7 MB, 236.4 MB smaller than YOLOv4. The confidence of the YOLOv5 network model is 0.91, 0.99 and 0.95 respectively when detecting the boat tractor, man and farm tools. The depth estimation of YOLOv5+SGBM within 2 m, and the accuracy reaches 98.1%. The paddy field depth estimation based on YOLOv5 and SGBM can meet the actual requirements of unmanned boat tractor with rotary tillage.

Aiming at the problem that traditional detection networks are difficult to extract the feature information of crop leaf pest and disease accurately and efficiently, a multilevel and multiscale feature fusion recognition algorithm for crop leaf pest is proposed through the improvement of YOLOv8 network. Firstly, a multilevel feature coding module is constructed to learn the comprehensive feature representation by learning the direct feature relationships of different levels of features. Then, a multiscale spacechannel attention module is designed on the basis of Transformer to capture the complementary relationships between different scales of features by learning the comprehensive multiscale feature representation patterns such as finegrained and coarsegrained, and all feature representations are effectively. The whole feature representations are fused, and the better recognition results are obtained.Finally, the experimental validation is conducted on the Plant Village public dataset, and the results show that the proposed improved method can effectively improve the alignment accuracy and accurately recognize different pests and diseases existing on the leaves of crops at the same time, and the mAP 0.5 for tomato leaves detection reaches 88.74%, which is 8.53% higher than the traditional YOLOv8 method, without significant increase in computation time. The ablation experiments also fully demonstrate the effectiveness of the proposed modules, which can better achieve highprecision leaf insect and disease recognition and provide a strong support and guarantee for the intelligent management of farmland.

In order to solve the problems that the characteristic parts of insect pests in the early stage of lemon were too small and difficult to detect along with the limited data sets, a YOLOv5-DK detection algorithm was proposed. The algorithm was based on YOLOv5, and adopted K-Means++ to recluster the anchor frame of the initial pest location of lemon, alleviating the problem of small pest characteristics at the initial stage of lemon. Meanwhile, a new lightweight Denseneck-2 module was proposed, which applied the idea of reuse in the DenseNet network, so that  the input of each layer of the detection algorithm had the characteristic information of each layer in front of it, which decreased the YOLOv5-DK demand on the initial sample volume of insect pests of lemons. The new YOLOv5-DK detection algorithm demonstrated higher competencies than the original one, including an increase of 3.4% in the average accuracy of detection, a decrease of 2.1% in the missed detection rate, and a reduction of 6.3% in the number of parameters of the algorithm model. These results showed that the algorithm performed better in the application of small samples and small targets.

In order to study the automatic bagging technology for apples suitable for the current production practice and realize the accurate recognition of immature green apples in the real environment, this paper proposed an image recognition method of immature green apples in natural scenes based on the improved YOLOv3. Firstly, in order to improve the recognition accuracy of immature green apples in images containing interference factors, this paper was based on the idea that YOLOv3 algorithm utilized residual network and multiscale feature fusion for detecting small targets, and improved and experimentally verified the YOLOv3 feature extraction network by utilizing the feature maps with dimensions of (104, 104, 128) instead of the original feature map with dimensions of (13, 13, 1 024) as the output. The improved YOLOv3 target detection model for immature green apples was proposed, which improved the ability of the algorithm network to capture immature green apples in the image and the recognition accuracy by increasing the size of the output feature maps of the feature extraction network and decreasing the size of the receptive field. Secondly, this paper designed the recognition comparison test under different algorithms, different varieties and different environments, and compared and analyzed the results. The mean Average Precision and Recall of the improved YOLOv3 on the overall dataset were 92.46% and 87.6%, respectively, which were 3.22% and 14.57% higher than that of the original YOLOv3. The performance enhancement of the improved model was mainly reflected in the ability to detect the correct number of targets. The mean Average Precision of the improved YOLOv3 on the test set of images containing the effects of illumination, overlapping and occlusion effects was improved by 3.58% and 2.74% compared to the original YOLOv3, respectively. The improved YOLOv3 model had higher detection accuracy on the overall dataset and on the test set of images containing interference factors, higher number of correct targets detected, and better antiinterference ability.

 Hemerocallis citrina Baroni has a short picking cycle and relatively strict picking requirements. Aiming at the problems of  the low efficiency and high subjectivity of manual harvesting of Hemerocallis citrina Baroni, a deep learning-based SSH-YOLOv5 Hemerocallis citrina Baroni maturity detection algorithm was proposed. Based on the YOLOv5 model, combined with the lightweight network ShuffleNet V2 basic residual unit to compress the size of the network model, and improve the model target detection speed. The attention mechanism module of Squeeze-and-Excitation network was integrated into the model to enhance the sensitivity of the model to useful feature information, and improve target detection precision, and ordinary convolution was replaced with depthseparable convolution module to further reduce the model computation. The experimental results showed that the number of parameters and floating point operations of the improved SSH-YOLOv5 model were reduced by 61.6% and 68.3% respectively, and the number of network layers was reduced by 18%, while the detection precision of SSH-YOLOv5 was improved from 88.8% to 91.2% of the original algorithm. The real-time detection speed reached 66.4 f/s, which was 18.1% higher than the original YOLOv5 algorithm and met the real-time detection requirements. The improved algorithm not only makes the model lightweight, but also makes Hemerocallis citrina Baroni maturity detection more accurate and faster, which can better meet the demand of Hemerocallis citrina Baroni detection.

The traditional target detection algorithm can only get the target frame, and cannot determine the growth direction of daylily. Aiming at this problem, the neural network structure is optimized on the basis of the existing target detection algorithm, and the prediction of the detection box is changed to the prediction of the key points. Firstly, the growth direction and length of daylily are determined according to the anchor point matching method, and the growth angle and length of the daylily target are counted. Multiple anchor points are set based on the statistical results. The actual growth angle and length are compared with the anchor points to obtain the relative length and angle of the target, which is used as the model prediction value for training. Secondly, the heat map prediction branch is added to the model to predict the four key points. Finally, the growth posture of daylily target is obtained by using the target length and angle information to connect the key points. An evaluation model method for line segment fitting characteristics is designed, Introduction of Partial Affinity Fields in Calculation Accuracy, and the NonMaximum Suppression algorithm is improved accordingly. Through experimental verification, the accuracy of picking target recognition is 91.02%, the positioning accuracy is 99.8%.

The existing insect related recognition algorithms have few kinds of recognition, lack of classification and recognition algorithms for a large number of invasive insects, and are difficult to provide stable and efficient technical support for the recognition function of the integrated invasive insect system. In this study, 31 kinds of invasive insect images are collected, processed and divided into data sets. Based on four convolutional neural network models, DenseNet121, MobileNetV3, ResNet101 and Shuffle Net, training, testing, analysis and discussion are carried out. The results show that MobileNetV3 has better comprehensive performance in the background algorithm application of the identification function of the invasive insect integrated recognition system. According to the existing defects and model characteristics of the MobileNetV3 model, the attention mechanism and activation function of the designated bottleneck layer of the MobileNetV3 model are improved. The accuracy of the improved model is 92.8%, and the average recognition time of a single test set image is 0.012 s, which is 0.5% higher and 15.2% shorter than the original MobileNetV3 model, which can well meet the requirements of multi insect recognition and classification.

The path tracking accuracy of intelligent agricultural machinery is affected by many factors such as field terrain and machinery structure, which may lead to serious consequences such as crushing the drip irrigation belt. In this paper, taking the cotton planter as the research object, a switching tracking algorithm consisting of Stanley tracking algorithm and linear quadratic optimal control (LQR) algorithm is proposed to guide the agricultural machine to enter the line quickly and maintain the straight line accuracy. In addition, in order to eliminate the static lateral error, an extended Kalman filter (EKF) for heading angle error is added. The simulation results show that the switching tracking algorithm can eliminate the static lateral error after entering the straight target line. The real vehicle test results show that when the given initial lateral error is 0.5 m and the speed is 3.6 km/h, the entry time is 6.88 s, and the overshoot is 0.041 m. When the given initial lateral error is 0 m, and the speed is 3.6 km/h, the straightline tracking accuracy is controlled within ±0.025 m, which meets the requirements of highprecision straightline operation of actual agricultural machinery. It shows that the algorithm studied has good tracking accuracy and antiinterference ability and it is conducive to improving agricultural production efficiency.

In order to accurately detect citrus infected by Bactrocera dorsalis (Hendel) using visible/nearinfrared spectroscopy, this study designed a multipath nondestructive detection grading system for citrus infected by Bactrocera dorsalis (Hendel) to address the unknown location of infection. Semitransmissive spectral information from four detection paths was collected and partial least squares discriminant analysis was used to establish and compare classification models for a single detection path and a combination of four detection paths. The results showed that the model combining all detection paths achieved better classification results. Using 47 feature wavelengths selected by the competitive adaptive reweighted sampling method to build a model, the accuracy and specificity of the best prediction set among the four detection paths reached 93.5% and 95.2%, respectively. The PLSDA hybrid classification model established by using citrus samples from four detection paths, combined with the CARS algorithm for effective feature wavelength variable selection, can improve the accuracy of Bactrocera dorsalis (Hendel) infected citrus classification model and accurately classify Bactrocera dorsalis (Hendel) infected citrus. The research results can provide a reference for online detection of citrus infestation by Bactrocera dorsalis (Hendel).

In the process of injecting the virus into the egg embryo to culture the vaccine, the impact of the needle caused cracks around the injection hole and  the culture failed. In order to solve the problem of low efficiency and high misjudgment rate in the detection of egg embryo cracks by the naked eye in the darkroom, a method for detecting cracks in egg embryos with holes combined with a multispectral channel attention mechanism (MSCA) and parallel stacking topology modules (PST) was proposed. Firstly, a black box for crack detection of perforated egg embryos was built, and the images of linear cracks, mesh cracks, starshaped cracks and intact egg embryos with holes after virus injection was collected and a data set was established, then ResNet-50 was used as the backbone model. The first residual module of the next 4 layers was replaced by the PST module to increase the image expression ability of the model at the beginning, finally, the MSCA mechanism was introduced after each PST module and the residual module, and the MSCA mechanism compressed the data through twodimensional discrete cosine transform (2D DCT),  and the neural architecture search selection (NAS) method obtained the optimal frequency components, which could redistribute the weights to increase the proportion of crack feature weights. Thus the microporous egg embryos could be identified quickly and accurately by the model. The results show that the improved network model has a detection time of 0.42 s/piece for cracks in egg embryos with holes, a detection accuracy of 96.43%, and a higher detection efficiency than manual operations. Compared with the original ResNet-50, the detection accuracy has been improved by 3.66%, which is superior to other classical convolutional network models. It is proved that the improved model can be used for the automatic detection of cracks in egg embryos with holes before vaccine cultivation.

Aiming at the problems of low accuracy of tomato disease identification due to overlapping leaves and small targets in complex environments, a multiscale cascade model (IMS-Cascade) is proposed. The model is based on cascade neural network (Cascade R-CNN), the switchable Atrous convolution of fused context information is introduced into the backbone network, and complex multiscale convolution kernels are used to extract target features to solve the problem that the shape of the same disease is greatly different due to leaf occlusion, and the feedback connection module is added to the feature pyramid networks, so that the model can extract features for many times and improve the utilization of shallow information. Finally, the gradient of accurate samples is increased in the loss function to reduce the influence of abnormal samples on the model. When the model is applied to a portion of the tomato leaf disease dataset published by Plant Village, the mean average precision (mAP) reaches 89.1% and the average precision reaches 99.36%. These results represent improvements of 2.5% and 1.84%, respectively, over the original Cascade R-CNN model. This indicates higher detection accuracy, which is beneficial for tomato disease detection in complex environments.

 In order to judge the fall of soybean flowers during the flowering period, soybean flowers in the field were accurately detected under four growth states such as flower bud, halfopening, fullopening and withering. Based on the YOLOv5 detection model, the backbone Bottleneck CSP structure was modified, the number of modules was reduced to preserve more shallow features and enhance feature expression ability. CA attention mechanism was introduced into the backbone network to obtain location information and help the model identify more accurately. Moreover, the size of anchor box was modified to improve the accurate identification of small target bud, and the improved YOLOv5 algorithm was compared with the selfbuilt data set of different growth states of soybean flowers in the field. The results showed that the accuracy rate of the model reached 93.4% and the recall rate reached 91.4%, which were increased by 0.8% and 2.1% respectively compared with the original model.

Aiming at the problem that the existing methods of crop disease leaf detection were not accurate enough to locate the leaf disease region by using image features,  a new method of crop disease leaf detection based on multimodal feature alignment was proposed. During the training phase, image and text from a collection of crop leaves were first encoded using visual and text encoders. The diseased areas in a given image were located according to the visual encoding features, and the integration of visual and text encoding features was used to achieve finegrained classification of the type of disease in the diseased area. In the inference phase, the pretrained disease area localization module was used to locate the diseased areas in a given test image, and the extracted diseased areas were used as input for a pretrained classification model. Finally, by calculating the similarity between the predicted text values and the original labels in the text set, a rapid finegrained classification result for the diseased area was obtained. Tests on several opensource crop disease datasets show that the proposed method can achieve high precision rates of 0.957 4, 0.961 1, 0.958 0, and 0.950 2 on potato, tomato, apple, and strawberry datasets, respectively. It has better comprehensive perfor mance and good paratical application value.

Rapid localization and accurate identification of tomato leaf diseases can help in the rational use of pesticides, thereby ensuring the quality and yield of tomatoes. In order to address the problem of poor performance of existing detection methods for tomato leaf disease, a selfsupervised detection method  for small sample tomato leaf disease was proposed. Firstly, a set of shared weight backbone networks were used to extract semantic features of tomato leaves in the visual space. Then, the visual semantic features were input into a deep autoencoder network, and the feature encoding network was optimized by calculating the contrast loss between the encoded and original features. Finally, the encoded features were used to guide the localization and identification of unknown tomato leaf diseases. In addition, a double loss optimization strategy was designed to obtain more robust guiding feature sets. Through testing experiments on a selfbuilt tomato disease leaf dataset and an opensource dataset, the proposed model achieved recognition accuracies of 0.946 2 and 0.963 9 on the selfbuilt and opensource datasets, respectively, which were superior to current stateoftheart object detection methods.

In order to provide reliable insect population data for more accurate application, a method of calculating the number of Laodelphax Striatellus (Fallen) based on image recognition technology was proposed. Three groups of shooting condition were set, which were composed of different shooting distances and camera adjustment factors. A recognition and counting model was formed by the combination of three parameters  of the region area, the region roundness and the boundary diameter. The edge detection and region filling were used to complete the extraction of individuals in images, and the parameters values of five single long wing images and five single short wing images were calculated separately in each group condition as standard. Under the same shooting conditions, 4 independent and 4 slightly connected Laodelphax Striatellus (Fallen) count images were calculated respectively, and the results of each region were compared with the standard value range of parameters. If the three parameters were all within the range, 1 was output; If at least one parameter did not match, recalculate the area of the region and the boundary diameter, and output 2 when it met 2 times the standard value at the same time. The results showed that 8 of the independent images had a relative error rate of 0%, 3 were less than 10%, and 1 was 13.3%. Among the slightly connected images, the relative error rate of 2 was 0%, 6 was less than 10%, and 4 was 10%-25%, which could satisfy the automatic calculation of the number of Laodelphax Striatellus (Fallen).

In order to improve the operating stability of the corn harvester engine in service, the engine operation data is taken as the analysis object, and a stability operation trend curve of the starter is calculated to judge whether the engine is in a stable state. Firstly, by obtaining operational data from the engine ECU system, the data under normal engine operation is normalized. Then, a prediction model is established based on a BP neural network optimized by genetic algorithm. Predicting the numerical value of oil pressure using four parameters: engine speed, agricultural machinery speed, coolant temperature, and system voltage as inputs, and the final decision coefficient of the prediction model reaches 0.88, which proves that the prediction model has a high degree of fit and can accurately predict the engine oil pressure. Under normal operation of the engine, the predicted deviation of oil pressure is relatively small. The benchmark vector set is constructed by combining the residual of a large number of normal operation oil pressure predicted values and actual operation values with the actual operation values, and the evaluation vector is constructed by combining the residual of 20 000 normal operation oil pressure predicted values and actual operation values with the actual operation values. The distance value between each evaluation vector and the reference vector set is calculated by using Markov distance. This distance value can represent a stability index value under the normal operation state of the engine. The analysis results show that the 20 000 index values obtained have a certain aggregation, and the index value is stable between 0 and 10. Therefore, a trend curve of this index value in the time series can represent a stable trend under the service state of the engine, and can be used to judge whether the engine is in normal or abnormal state.

It is an important indicator for the design of the suspension system of tractors in hilly and mountainous areas to ensure the optimal matching state between traction and plowing depth when the tractor is working on hilly terrain or hard soil. Proportional electromagnets is the main component of throttle valve electromechanical conversion,  its performance is of great significance in controlling throttle opening and achieving proportional adjustment of flow rate. This article focuses on the operational requirements of the suspension system of tractors in hilly and mountainous areas and the design requirements of the hydraulic proportional valve, a pilot throttle valve proportional solenoid for adjusting the attitude of the suspension hydraulic system is designed, and the key structural parameters of proportional solenoid are calculated. A mathematical model of the solenoid is established, and the impact of Maxwell on the flow characteristics of the valve group is simulated and analyzed. At the same time, genetic algorithm was used to optimize parameters such as the thickness of the magnetic isolation guide sleeve, the angle of the magnetic isolation ring, the width of the magnetic isolation ring, and the depth of the stator core. Finally, Maxwell was used to conduct finite element analysis of the optimized proportional electromagnet before and after optimization. The simulation results revealed that the electromagnetic force of the optimized proportional electromagnet was almost parallel to the Xaxis between 1-3 mm, and had better displacement force characteristics. The mathematical model of the pilot throttle valve was established, and its dynamic and static characteristics were simulated by Simulink so as to verify the rationality of the designed proportional solenoid parameters.

In order to obtain the accurate discrete element parameters of tobacco nutrient soil, the simulation parameter calibration of tobacco nutrient soil discrete element was carried out by combining simulation test and bench test. Taking the Angle of repose of tobacco vegetative soil as the response index and the Hertz-Mindli (no slip) discrete element contact model was used to design the Plackett-Burman test, to screen the discrete element simulation parameters of tobacco vegetative soil. The factors that had significant influence on the  resting Angle of tobacco vegetative soil were the static friction coefficient between vegetative soil, the rolling friction coefficient between vegetative soil and the static friction coefficient between vegetative soil and glass. In order to further obtain the optimal parameter combination of each factor, Box-Behnken experiment with three factors and three levels was designed, and the secondorder regression response model of the restangle parameters of tobacco vegetative soil was obtained. The results showed that the optimal parameter combination of discrete element model of tobacco nutrient soil was as follows: static friction coefficient of 0.67, rolling friction coefficient of 0.35, static friction coefficient of vegetative soil and glass of 0.35. Under the discrete element model obtained from calibration, the relative error of the resting  Angle obtained by simulation test and bench test is 2.59%.

In view of the problems of low removal efficiency and easy damage of pepper mechanized processing, the physical and mechanical characteristics of pepper are tested. In this paper, with Xinjiang Red dragon pepper as the research object, the main physical parameters and their correlation were obtained through physical characteristic parameter measurement and statistical data processing. The length of pepper was 141.00 mm, diameter of pepper was 18.35 mm, length of pepper stalk was 31.90 mm, diameter of pepper stalk was 5.97 mm, and single fruit weight of pepper was 5.64 g. With the help of shear and tensile mechanical tests, the mechanical properties of pepper stalk and the joint between pepper stalk and pepper pedicle were obtained. In the shear test, the average shear strength of the pepper stalk was 9.73 MPa, through the Design-Expert 12 software optimization design, the minimum shear strength was 6.34 MPa at the shear position, stalk diameter and loading speed were 15 mm, 4.5 mm, and 200 mm/min, respectively. In the tensile test, the average tensile strength of the pepper stalk was 5.13 MPa, through the Design-Expert 12 software optimization design, the minimum tensile strength was 3.95 MPa at the grip position, stalk diameter and loading speed were 55 mm, 4.4 mm, and 395.5 mm/min, respectively.

 In order to realize the popularization and application of simple preservation and storage equipment, it is necessary to determine the quality evaluation method of simple preservation and storage equipment. By analyzing and studying the quality investigation and evaluation method of simple preservation and storage equipment and carrying out practical application, the safety, reliability, adaptability and after-sales service status of simple preservation and storage equipment are investigated and evaluated, and the overall quality level of in-use preservation and storage equipment is mastered. Based on the analysis and study of the composition, parameters and functional principle of simple preservation and storage equipment, the possible conditions and faults in the actual operation and use of the equipment, the quality requirements of the user's use of the equipment and the service experience, the quality evaluation indicators of the equipment are set up, and each indicator is assigned a different weight coefficient. Among them, there are 5 first-level indicators, which are safety, applicability, reliability, material use load degree, after-sales service, and there are 25 secondary indicators. In the first level index, safety 25 points, applicability 20 points, reliability 20 points, material usage load 20 points, after-sales service 15 points. The results show that the weighted score of all samples of simple fresh-keeping storage equipment is 83.10. The production performance of the products of the three companies is good, and the product design standards are reached.

 In the face of the increasingly large supply market of agricultural products, the serious centralized structure of the existing traceability system, fuzzy and simple traceability information, and difficult to effectively guarantee data security and other problems, by analyzing the business process and related characteristics of the supply chain, this paper integrates blockchain and IPFS technology to design a flexible multi-chain distributed traceability system to meet the needs of accurate traceability and supervision in the complex and changing market situation. It makes full use of the deduplication storage and distributed storage characteristics of Interplanetary File System (IPFS) to improve the utilization of storage resources, and solves the problems of severe centralization, single data type, and excessive load of traceable data storage in the past. The combination of blockchain multi-chain and IPFS private network and the adoption of CBC and ECC encryption algorithms can support comprehensive market regulation, and can meet the encryption protection needs of enterprise privacy data. Finally, in order to verify the validity of the model, the hydroponic vegetable traceability project is used as an application case to analyze and test, in which the throughput of the blockchain multichain system has reached 500 TPS for the upchain and 400 TPS for the query. The ciphertext change rate of CBC encryption algorithm has reached 96.727% and 97.136%, respectively, in correlation and expansibility analysis. The test results show that on the basis of the realization of efficient supply chain traceability data record store and accurate supervision and traceability,  the traceability system can also give the parties' legitimate rights and interests within the chain and the effective protection of privacy data, can also face the business crisscross and complex market supply chain, providing reference for the research, development and implementation of   a complete and reliable blockchain of agricultural products traceability supervision system.